Voted #1 \"Best Places to Work\" by the Phoenix Business Journal, Achieve is a leader in the digital personal finance industry built to help everyday people thrive

Achieve

achieve

The TechBeat: Backing Up Weaviate with MinIO S3 Buckets to Achieve Strategic Enhancement to Data Management (4/13/2024)

The TechBeat: Backing Up Weaviate with MinIO S3 Buckets to Achieve Strategic Enhancement to Data Management (4/11/2024)

Backing Up Weaviate with MinIO S3 Buckets to Achieve Strategic Enhancement to Data Management

ডাটা ম্যানেজমেন্টে কৌশলগত বর্ধন অর্জনের জন্য MinIO S3 বাকেটের সাহায্যে ওয়েভিয়েট ব্যাক আপ করা হচ্ছে

Sichern von Weaviate mit MinIO S3-Buckets zur strategischen Verbesserung des Datenmanagements

Realizar una copia de seguridad de Weaviate con MinIO S3 Buckets para lograr una mejora estratégica en la gestión de datos

Sauvegarde de Weaviate avec les buckets MinIO S3 pour obtenir une amélioration stratégique de la gestion des données

डेटा प्रबंधन में रणनीतिक वृद्धि हासिल करने के लिए मिनियो एस3 बकेट के साथ वेविएट का बैकअप लेना

MinIO S3 バケットを使用して Weaviate をバックアップし、データ管理の戦略的な強化を実現

데이터 관리의 전략적 향상을 위해 MinIO S3 버킷으로 Weaviate 백업

Fazendo backup do Weaviate com buckets MinIO S3 para obter aprimoramento estratégico no gerenciamento de dados

Резервное копирование Weaviate с помощью сегментов MinIO S3 для достижения стратегического улучшения управления данными

Veri Yönetiminde Stratejik İyileştirme Sağlamak için Weaviate'i MinIO S3 Kovalarıyla Yedekleme

Sao lưu Weaviate bằng nhóm MinIO S3 để đạt được cải tiến chiến lược trong quản lý dữ liệu

使用 MinIO S3 Buckets 备份 Weaviate，实现数据管理的战略增强

In the presentation below, 0xGorilla, CTO at , presented on the current state of price oracles and the steps to improving them. DeFi Wonderland https://youtu.be/HR9rRp6c7Y8?si=Q1DhUBDXGOvzTmHZ&embedable=true Below is a glossary of key concepts mentioned during 0xGorilla’s talk, intended as a supplement to his video presentation. WHAT IS DEFI WONDERLAND? is a community of developers, researchers, investors, and other DeFi enthusiasts with the collective mission of fostering a decentralized and open financial landscape. The team works with prominent DeFi protocols such as , , and to push the boundaries of innovation in the DeFi space. DeFi Wonderland Optimism Reflexer Connext PRICE ORACLE is a permissionless oracle solution built on Uniswap V3 to provide reliable price quotes on any existing token. The project reduces complexity for users and protocols and removes the trust issues associated with liquidity providers. Price aims to address the oracle problem by leveraging Uniswap’s decentralization and sustainability while using automation to ensure reliability. Price Oracle CARDINALITY ARRAY Cardinality arrays are data structures used by price oracles to represent and process multiple price data points from various sources. They enable oracles to efficiently collect and process data, simplify the calculation of Time-Weighted Average Prices (TWAP), and help in the application of statistical analysis on the price history. To use cardinality arrays, oracles collect data with each element representing a single price data point e.g., the price of an asset at a specific timestamp. Next, the arrays store the data efficiently and the price oracles can now process them into accurate information for smart contracts and dapps. Additionally, cardinality arrays may be synchronized to prevent data inconsistencies. TWAP (Time-Weighted Average Price) QUERYING TWAP is the average price of an asset over a specific period, which is adopted by DeFi protocols to avoid price manipulation. TWAP querying refers to the process which provides the TWAP price through average price calculations using data from an oracle or data source. The first step in TWAP querying is to select the time interval over which the average price will be calculated and then collect price data for the asset at regular intervals during that period. After this, the oracle calculates the weighted average price and performs data validation to check for outliers, data inconsistencies, and potential manipulation attempts. Finally, the Time-Weighted Average Price is obtained as output. SPOT PRICE Spot price represents the immediate value of an asset in the market without accounting for any future changes. In DeFi where prices are volatile and prone to manipulation, oracles avoid returning the current spot price. Instead, oracles use price-averaging methods like Time-Weighted Average Price (TWAP), Volume-Weighted Average Price (VWAP), Simple Moving Average (SMA), and many others to provide stable and manipulation-resistant reference prices. More information about TWAP and VWAP algorithms can be found in by Chainlink. this article Learn more about DeFi Wonderland here: Wonderland Website Wonderland GitHub Wonderland Twitter 0xGorilla Twitter Price Oracle Documentation The Blockchain Oracle Summit is the world’s only technical summit that dives deep into the use cases, limitations, and impacts of oracles on the wider blockchain ecosystem. Leading speakers worldwide gathered in Paris to share their work and experience building and using oracle solutions. Article by . Michael Abiodun Also appears . here

Case Study: How To Achieve A Good On-Chain Solution For Price Feeds Post Merge

কেস স্টাডি: একত্রিত হওয়ার পরে মূল্য ফিডের জন্য একটি ভাল অন-চেইন সমাধান কীভাবে অর্জন করা যায়

Fallstudie: So erreichen Sie eine gute On-Chain-Lösung für Preis-Feeds nach der Zusammenführung

Estudio de caso: Cómo lograr una buena solución en cadena para los feeds de precios después de la fusión

Étude de cas : Comment obtenir une bonne solution en chaîne pour les flux de prix après la fusion

केस स्टडी: मर्ज के बाद मूल्य फ़ीड के लिए एक अच्छा ऑन-चेन समाधान कैसे प्राप्त करें

ケーススタディ: マージ後の価格フィードに対して優れたオンチェーン ソリューションを実現する方法

사례 연구: 병합 후 가격 피드에 대한 우수한 온체인 솔루션을 달성하는 방법

Estudo de caso: como conseguir uma boa solução on-chain para feeds de preços pós-mesclagem

Практический пример: как создать хорошее внутрисетевое решение для ценовых потоков после слияния

Örnek Olay: Birleşme Sonrası Fiyat Akışları İçin İyi Bir Zincir İçi Çözüm Nasıl Elde Edilir?

Nghiên cứu điển hình: Cách đạt được giải pháp tốt trên chuỗi cho nguồn cấp dữ liệu giá sau khi hợp nhất

案例研究：如何为合并后的价格反馈实现良好的链上解决方案

Process mining isn't a novel concept. Since its emergence in 2011, process mining has been undergoing continuous improvement as leading providers like and consistently refine and expand their suite of tools. Celonis UiPath The process mining market is expected to reach , so understanding both the direct and hidden benefits of the methodology is crucial for enterprises aiming for a competitive edge. Therefore, it makes sense to break down the value and purpose of this technology, while outlining the key steps for achieving the most rewarding results. by 2028 $12.1 billion The definition of process mining Process mining involves extracting and utilizing data to explore and improve enterprise workflows. The reason why it is important is directly connected to the process most common reason for failed digital transformations— , as noted by Organizations resist change when they can’t see the entire journey clearly and thus fail to align with leaders' thoughts. lack of vision and direction Deloitte. Business leaders may not provide specifics, resulting in unclear transformation goals, undefined timeframes, and a lack of transparency for employee motivation within updated workflows. Low visibility arises from the abundance of enterprise data and processes requiring detailed sorting and examination. However, manual handling is time-consuming, error-prone, and leads to blind spots rather than insights. Due to this, process mining becomes a game-changer by collecting log event data, transforming it into a graph that visualizes a specific enterprise process and its cycle. By using process mining technology, executives and decision-makers can and gain insights on potential areas of improvement. view process performance in different scenarios Process mining value Process mining offers new opportunities across the enterprise and brings efficiency to various departments Marketing and sales productivity boost Process mining allows breaking down sales processes, identifying weak points. It also delivers insights into lead and client interactions. Monitors sales process: visualizing processes and interactions for KPI tracking. Designs agile sales process: defining success criteria for a flexible process. Increases conversion rates: identifying and addresses stages limiting conversion. Optimizes customer experience: defining activities impacting customer choices. Improves marketing strategies: evaluating lead responses to marketing campaigns. Compliance and auditing facilitation According to an EY survey, process mining can optimize control-related activities, including auditing and regulatory compliance, in about of cases with: 19% Accelerated issue discovery: uncovering the reasons causing operational issues through detailed workflow visualization. Transparent financial reporting: analyzing event logs for financial transactions. Improved financial control testing: identifying control problems and non-compliance issues for consistent auditing instead of periodic assessments. Cost reduction: reducing manual tasks and optimizing employee time management. How to get max value from process mining Over 12 years, process mining has evolved from a concept to a core technology embraced by vendors like Celonis and UiPath. As new approaches emerge, the landscape of implementing process mining technology Despite this, executives and decision-makers should focus on the following essentials. will see changes. Form a team Process mining was once a collective effort, involving data gathering and employee surveys for . Now, it focuses on engaged personas instead of employees. actionable visualizations Executive: sets process mining project goals and secures funding and resources. Process owner: is responsible for end-to-end process execution and productivity, defining desired processes, and monitoring. Project manager: handles all aspects of the process mining task, selects technology, manages team communication, and reports progress. Business analyst: analyzes process mining data, identifies opportunities and risks, proposes solutions, and monitors process changes. Data engineer: extracts, transforms, and merges log event data for creating process graphs. BI engineer: connects data with users, ensures a smooth user experience, and facilitates data visualization for process graphs. Infrastructure engineer: ensures proper integration of the process mining solution within the enterprise structure, considering security and compliance. Check data system health For any data-driven process to succeed, having relevant, validated, and healthy data is essential. To ensure fruitful process mining outcomes, executives and project managers must verify the availability of necessary event data and sufficient data history. Identify the need A successful process transformation begins by recognizing a clear need. To avoid blind spots, business leaders should thoroughly investigate the issue prompting the idea for process automation, considering various levels. The depth of research directly impacts the final results and process mining goals. Establish KPIs After understanding enterprise needs, executives and stakeholders move on to the crucial step of defining primary key performance indicators. It involves specifying data sources for insights extraction, considering data dimensions (region, supplier, etc.), and estimating costs and expected outcomes. Develop process maps For successful transformation, creating detailed and multilayered process maps is essential. These maps visually represent the process, facilitating conformance checks to find deviations, process violations, and bottlenecks. Act on insights Contrary to the belief that process mining is only for insight-gathering, it can also contribute to immediate improvements. After identifying opportunities , it's beneficial to address issues promptly instead of postponing for later stages or automation. through process graphs and checks For example, process owners and analysts can streamline processes before implementing intelligent automation, ensuring a smoother transformation. Addressing issues upfront enhances the success of subsequent automation efforts. How to secure consistent growth Intelligent automation is a continuous process that constantly demands insights and awareness of enterprise operations. To improve process mining productivity, there are several important guidelines for business leaders to follow. Decision-makers should regularly update their knowledge of process mining and intelligence trends, connecting with peers and consulting digital partners for valuable insights. Stay updated on trends. Promoting innovation understanding among employees reduces resistance to change. Leaders should consistently assess employees' perceptions of transformation, ensuring friction-free initiatives with clear, trackable results. Foster innovation culture. Maintaining data health by preventing bottlenecks and silos. Investing in advanced data validation methods, and constantly updating and improving security measures for smooth integration across systems. Prioritize data hygiene.

How to Achieve Optimal ROI Through Process Mining

A couple of days ago, I wanted to invest a certain amount into a basket of stocks. But, I encountered a dilemma on how much to invest in individual stocks. I aim to maintain a target allocation ratio for each stock to have a balanced portfolio. Consider the example that you have a portfolio of 3 stocks viz. INFOSYS, TCS and WIPRO. Your current investment in each of these stocks is ₹30, ₹50, and ₹20 respectively (currency is not important here as long as all stocks are in the same currency). Let’s say our Target Allocation ratio (TAR) for each of these stocks is 33%, 33%, and 34% respectively as shown in the following table. Instrument Current Investment Current TAR Target TAR INFOSYS ₹30 30% 33% TCS ₹50 50% 33% WIPRO ₹20 20% 34% Total ₹100 Now, what’s the optimal choice if you want to invest a total of ₹30 to get closer to your TAR for each of these stocks? I googled for some tools but couldn't find any that could help my case. So, I sat in to derive an equation that could provide me with an optimal allocation. Mathematics You have a portfolio of financial instruments: n and your current investments in each financial instrument is: This means that your current investment allocation ratio (between and ) is: 0 1 Now, suppose you want to invest an additional amount of split between these financial instruments. You want to achieve a target allocation ratio of the following for each of these instruments. SN n This implies that: And, our objective is to find out these i.e. how much investment to make in each of the financial instruments. N_i n To solve this problem, we have converted it into an optimization problem. If we assume some initial allocation for each instrument, we calculate the Mean Squared Error between the objective allocation ratio and the actual allocation ratio. Then, we can simply use any (bounded and constrained) optimizing tool to solve the problem. MSE So, Now, we can simply minimize this function to find all N_i Assumptions No selling. The whole point of writing this code is that we don't want to sell any existing instruments to achieve the target allocation. We assume that we would want to invest the full amount of i.e. if you input ₹100 as a new investment to be allocated, the model would allocate the full amount between the financial instruments. new_investment We don't solve the above problem rigorously. The correct way would be to create equations for variables and solve them using an equation solver. However, if that were easy, we wouldn't have invented Machine Learning. So, we instead solve the problem by minimizing the between target and actual allocation ratio using a minimizer function (similar to how Gradient Descent would minimize the error function in Machine Learning). n n MSE Implementation I have converted the above problem into a simple GitHub project. You can find the same . The file details how to use the same. Essentially, I used to minimize the function above. here README scipy.optimizers.minimize MSE To showcase the same, you can edit the file to try out the example that we took in the beginning. src/views/file.py # list all financial instruments
FINANCIAL_INSTRUMENT_NAMES: list[str] = ["INFOSYS", "TCS", "WIPRO"]

# Mention initial investment in each instrument in your currency (all should be in same units)
FINANCIAL_INSTRUMENTS_INITIAL_INVESTMENT: list[float] = [30, 50, 20]

# Mention new investment to be made in same unit as initial investment
NEW_INVESTMENT: float = 30

# Mention TAR for each instrument that you want to achieve
FINANCIAL_INSTRUMENTS_TARGET_INVESTMENT_RATIO: list[float] = [0.33, 0.33, 0.34] Once you have edited the above values in , you can run the code by running the command . It would yield something like the following: src/views/file.py python -m src.views.file Optimizing the MSE to compute optimal allocation. Sit tight!
Final allocation of investment of unit 30 is as follows:
Instrument      Initial Investment %    Target Investment %    Suggested Investment    Final %
------------  ----------------------  ---------------------  ----------------------  ---------
INFOSYS                           30                     33                      10    30.7692
TCS                               50                     33                      10    46.1538
WIPRO                             20                     34                      10    23.0769
Please manually verify the results as Mathematics can only aid you but can't absolve you! As we can see, the model recommended allocation of ₹10 for each instrument. Note that with the given investment amount, it might not be possible to achieve exact TAR but the model would help to move towards it. In the above case, has moved from to . has come down from to and has moved from to . INFOSYS 30% 30.7% TCS 50% 46% WIPRO 20% 23% Feel free to try out different values and let me know if you have any comments. Lead Image generated using Adobe Express

How To Achieve Target Allocation Ratio (TAR) in Your Investment Portfolio

$DEFI Token secures listings on the 7 most popular crypto exchanges, including ByBit, KuCoin, Gate, MEXC, Bitget & Bitmart. Trading will start on January 31st. DUBAI, the United Arab Emirates (UAE), January 30th, 2024/Chainwire/-- De.Fi, the Largest Web3 Super App, excelled with the most prominent public sale in 2023, raising a staggering $17M. Today, the team unveiled the thrilling news to their dedicated community. The highly anticipated $DEFI token is set to make its debut on seven top-tier exchanges, including , , , , , and . The listing event marks a significant milestone for the platform, which is set to redefine the future of DeFi. ByBit KuCoin Gate MEXC Bitget Bitmart DeFi aims to become the first Web3 service to launch a token. Currently, the team is entirely focused on ensuring a successful listing campaign. The dedication to this initiative reflects a commitment to creating a substantial and enduring presence for the $DEFI token in the market. According to the , users should expect the $DEFI listing on crypto exchanges on January 31st, 2024, at 10 AM UTC. official announcement The De.Fi team is backed by a range of leading Venture Capital firms and prominent investors from around the world, including 21Shares, Shima Capital, HOF Capital, and other leading VCs. Prominent partnerships with CoinGecko, zkSync, LayerZero, Trust Wallet, and Hacken show De.Fi's dedication to becoming the ultimate SuperApp that makes Web3 more secure and delivers the best possible user experience. As the buzz and optimism returned to the industry, De.Fi is doubling down on building integrations and partnerships with crypto projects, exchanges, and blockchains. Notably, over 65 crypto projects worldwide have joined forces with De.Fi, leveraging its API for development initiatives. GeckoTerminal, AvaScan, Ankr, HoldStation, Fox Wallet, and many others are already utilizing Antivirus technology to protect their users from crypto scams. Last month De.Fi hit the absolute record of 10,000,000 API calls! Just a couple of weeks ago, De.Fi presented their with an epic prize fund of $100,000. The Quest offers tempting rewards, interesting tasks, and an immersive dive into the realm of gamification. This is something airdrop hunters have been anticipating for a long time. But only the most consistent participants will be rewarded. Quest 2 According to the renewed roadmap, De.Fi has ambitious plans for this year, including the release of several new features: De.Fi Connect, a full-fledged Web3 Social Network for participants in the decentralized finance ecosystem De.Fi L2 Chain to scale and connect their Ecosystem De.Fi-Gpt, is a novel personal finance copilot that will help users make smarter crypto investment decisions. Additionally, after achieving a milestone of over 5 million users in 2023 with the launch of De.Fi 2.0, the team is poised to double this figure as part of their strategic growth plan. About De.Fi De.Fi is a Web 3 Super App because it offers a number of unique and valuable features. This all-in-one platform serves as a portfolio manager dashboard, facilitating tracking of crypto and DeFi positions across a large selection of chains, including non-EVM. Navigate through the Explore Yields feature to find the best DeFi opportunities effortlessly. The Crypto Antivirus, an innovative security system embedded in the dashboard, diligently alerts users to risky tokens and smart contracts. De.Fi SocialFi is gamifying the onboarding process of users into Web3, allowing them to create profiles, amass followers, build communities, earn points, and unlock rewards. Contact PR Manager Maria Lyshchuk De.Fi marketing@de.fi This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here.

$DEFI Token Hits 7 Major Exchanges: A Milestone Achievement

Bitcoin now has more than 250 million users at the last estimate. The reason I'm talking about estimates here is that Bitcoin is a pseudonymous network that gives users the ability to have multiple addresses on the network. We can, therefore, have precise on-chain data on the addresses active on the network, but we can only estimate the number of users behind these addresses. Having said that, it is important to bear in mind that while has 250 million users at the moment, only a tiny minority understand the why of Bitcoin. For a majority of investors in the Bitcoin world, Bitcoin is nothing more than an asset with a high potential return on investment. Bitcoin One of many, but mostly the one to own right now. For the few who understand the why of Bitcoin, Bitcoin is much more than that. Bitcoin is a once-in-a-lifetime chance to regain control of your money's future. Bitcoin gives you the freedom . to live your life on your own terms So, it is said that Bitcoin gives you self-sovereignty. However, this self-sovereignty is not instantaneous with Bitcoin. Just because you buy Bitcoin does not mean that you automatically take control over the fruits of your labor and, thus, ultimately, power over your life. To reach this stage of self-sovereignty, you will have to go through four levels in my opinion. You can, of course, go straight to the higher level 4, but generally, it is done progressively, step by step. As you understand why Bitcoin is there and how it works, you move towards that self-sovereignty. Level 1 - Buy Bitcoin The first step to self-sovereignty is obviously to buy Bitcoin. Many people only study Bitcoin or the economy but never take action. Taking action is the key to changing your life for the better. It's not about how much BTC you buy; it's about taking action. Start with the amount you feel most comfortable with, but don't stop there. Keep studying Bitcoin to understand how Bitcoin works and why Bitcoin exists. There is nothing worse than people who buy Bitcoin just to get rich and then never make the effort to understand the deeper meaning of its revolution. Bitcoin has the power to change your life for the better, but to do so, you will need to make the necessary intellectual effort that will give you the confidence to fully embrace its revolution. Level 2 - HODL Bitcoin Once you have acquired sufficient knowledge about Bitcoin, you will have confidence in its revolution. You will understand that Bitcoin is an accumulation game for years to come. Your interest is, therefore, to become a Bitcoin HODLer no matter what. By becoming a Bitcoin HODLer, you will stop worrying about the volatility of the Bitcoin price. This only affects traders. Bitcoin HODLers stay cool no matter what because they know that Bitcoin's goals are long-term. Bitcoin is an incredible monetary revolution. Its revolution will take time. Confidence in Bitcoin allows you to be patient. This is a good thing because Bitcoin always rewards patience. Level 3 - Own your private keys You probably already know the most famous slogan of Bitcoin: “ ”. This slogan is not misleading since Bitcoin allows you to verify everything yourself at any time. Don't Trust, Verify However, there is another slogan that is less well known to the general public that you will have to learn to reach this level 3 in the path of self-sovereignty: “Not your Keys, Not your Bitcoin.” You only own the Bitcoin whose associated private keys you have in your possession. When you buy BTC on a trading platform and leave that Bitcoin on the platform, you do not own that Bitcoin. The platform can potentially decide at any time to ban your account and confiscate your BTC. You fall into the same trap as with the current monetary and financial system in which banks can confiscate your assets at any time for arbitrary reasons. Your interest will, therefore, consist in transferring this BTC to cold storage. You will then take full control over your Bitcoin wealth. It's a daunting prospect for some, but you shouldn't be afraid to do this. It's much easier than some people would have you believe, and more importantly, it's the only way you can truly take responsibility for the security of the fruits of your labor. Level 4 - Run your Bitcoin Full Node After reaching level 3, which is already remarkable, many people stop there. This is a mistake, in my opinion. You have to go even further and become your own bank! . To become your own bank, you need to run a full node on the Bitcoin network By doing this, you will do even better, as you will become a full-fledged bank that relays transactions from the network. Some people say that there is no point for individuals to run a full node on the Bitcoin network. These people are wrong. By reaching level 4, you will be free to take full advantage of the trustlessness of Bitcoin. You won't have to believe the truth of others, but you can build your truth that is reinforced by the fact that it matches the truth of other nodes on the network. Solutions exist to run a full node on the Bitcoin network, and this is something that will allow you to reach the ultimate stage of self-sovereignty. This is what every Bitcoiner should strive for. Final Thoughts Bitcoin is much more than a financial investment with high potential returns. Bitcoin is an alternative system that will allow you to change your life for the better. To do this, you must fully embrace the Bitcoin revolution by taking the path of self-sovereignty. This path consists of 4 major levels. Your interest is to reach level 4 which will consist of buying Bitcoin, HODL Bitcoin no matter what while owning your private keys, and finally running a full node on the network. Then, the power will be in your possession. You will be able to live your life on your own terms. You'll have achieved self-sovereignty. Also published . here

Can You Achieve Self-Sovereignty With Bitcoin?: There are Four Levels to This

Last year, I wrote my first . I liked the experience, so I'm trying one more time. Let the future decide if it will become a trend or not. yearly retrospective Before diving into our safe technological world, my thoughts go to Ukraine, to my friends who had to flee their own country, to other friends who fought on the front to defend it from an imperial power, and to all victims of an old kleptocrat who clings to power despite the cost to others. The free world needs to support Ukraine more. I hope 2024 will be the year of Ukrainian victory. The AI revolution Last year, I kept the post focused on what I did. However, AI is pervasive in our tech world, if not the whole world, and deserves a dedicated section. Nobody even remotely connected to tech can ignore the buzz surrounding AI. Even friends and families who don't work near tech probably talk about it. So far, I didn't weave any AI-related thing into any of my talks, despite the huge incentive to do so: having AI in your Call For Paper proposal vastly increases your chances of being selected. And yet, it doesn't mean that I'm not playing with it on a personal level. Here's my experience so far. I've been playing with Dall-E as a use case to try out web development in Rust. I found the results mind-blowing. However, the limitations are enormous; it seems the size is always square, and the set of possible dimensions is limited. I've used ChatGPT in several areas. First, I tried to generate conference abstracts. I fed an entire blog post and asked ChatGPT to turn it into an abstract. Despite my lack of prompt engineering skills, I've found results severely lacking. Abstracts felt artificial, like any content created with ChatGPT, but worse, the abstract revealed too much or not enough. I also tried to use it to refactor two nested Kotlin statements, one nested in the other, to the null-safe construct. The first result didn't compile; the second one forgot a branch. It led me to the correct solution, though. if else let Finally, I recently started using GitHub Copilot as an extension inside of IntelliJ IDEA. Its behavior is fascinating: most of the time, it does nothing, but once in a while, it offers a snippet of a couple of lines, which is either entirely correct or at least very close to the target solution. Even better, the suggestions seem to be more frequent and even more relevant with time. All in all, I'm far from impressed by the current state of AI. However, I like Copilot a lot: I prefer rare advice, which is relevant, than the opposite. Technical content Next February will mark the two-year milestone that I'm working for: on . I'm still very pleased about both. It allows me to do things I like a lot, such as writing posts and giving talks. API7.ai Apache APISIX In 2023, I published fifty blog posts on : one each week on Sunday, but on Christmas and the New Year - for obvious reasons. Here are the top most viewed pages: this blog Rank Post Views Avg. time on page #1 Leverage the richness of HTTP status codes 14,848 :29 #2 A list of cache providers 12,770 1:15 #3 Calling Rust from Python 8,780 :52 #4 My final take on Gradle (vs. Maven) 8,238 2:11 #5 Learning by doing: An HTTP API with Rust 6,642 1:00 I continue to cross-post on different sites. Here are the numbers compared with last year's: Site 2023 2022 Medium 741 564 Dev.to 8156 1,838 Hashnode 89 80 Absolute numbers are not that interesting, but comparing them is. Interestingly, numbers on are growing wildly, while on Hashnode, they plateau. Note that other sites provide no precise follower count or no count. dev.to Besides, I created a script to track daily metrics across sites and social media, just as I did for Apache APISIX. It displays interesting results: As above, numbers are much less important than the trend. Growth visibly happens mainly on and Bluesky for reasons I cannot fathom (yet?). dev.to Finally, I started a weekly newsletter, unoriginally named A Java Geek weekly. So far, I've written a . I list the posts and videos I found interesting during the week. Note that they contain the same links I post on , , and , with a bit more content, either a summary or my opinion. couple of them LinkedIn Mastodon BlueSky Open Source contributions In 2023, after over twenty years in software, I finally became ! I'm both excited and impressed; it's like belonging to a group of mythical beings I'd only heard about. an Apache committer Of course, working on Apache APISIX helped a lot. Yet, the exciting bit is that all my contributions are entirely unrelated to code; they are blog posts, reviews of blog posts written by others, issues, comments on issues, etc. Here's the GitHub summary: Conclusion Last year, my resolutions were: To deepen my understanding of the Apache APISIX ecosystem Write as many blog posts as in 2022 - it's hard to do better, anyway Design at least three new talks I fulfilled the two first goals but unfortunately failed the last item. I have only created two new talks, both based on previous posts. I'll keep them secret for now; they should appear soon on my page in case they are selected. Speaking I lack imagination, so I'll keep the same goals for this year as last year's and add exploring the ecosystem as well. Let's see how it fares. Happy New Year! API7.ai Originally on on January 7th, 2024 published A Java Geek

Everything I Achieved in 2023

On the 1st day of 2022, I took a giant leap into a new career. Against limitations, I secured a job by the 10th month. The key to my success? Goal setting. It was the first time I stepped into a new year with daylight clarity on what I wanted to achieve and a roadmap for achieving it. But guess what? I didn't realize this until last December when I was reflecting on how I spent 2023. Despite achieving some feats, seeing how I shot my arrows aimlessly made me know I would've done better if I had a goal. With 2024 in sight, goal setting became a no-brainer. To do this, I demystified my 2022 goal to find the elements that made it work. I also drew inspiration from some successful entrepreneurs. All of which I'm excited to share with you in this article. So, keep reading if, like me, you believe 2023 would've been better with a clearer focus, and you're ready to rewrite the narrative in 2024. 9 Elements of an Effective New Year Goal 1. Choose the Right Goal Choosing a goal is the first and fundamental step in goal setting. And so, deserves more thought than it often gets. This is because the wrong goal will collapse any hard work you build on it. Just as the Bible states, "If the foundations be destroyed, what shall the righteous do?". The telltale sign of a wrong goal is when you choose it based on necessity. Such goals stem from the "I can't be left out/I can't appear unserious" mindset. They're fueled by societal expectations rather than intrinsic motivation. As such, it will make it hard for you to stay resolute when the journey gets murky. The best way to choose a goal is based on—Needs. Because they emanate from genuine reason, which increases your likelihood of committing to them. 4 Steps to Choosing a New Year Goal Based on Needs: Identify your Most Crucial Needs (MCN) Step 1: If you already know what they are/it is, skip to the next step. Assign them a mark on a scale of 10 to identify the parts of your life you're performing the least. And prioritizing the ones with the most significant impact on your current stage in life. Choosing your MCN is important to ensure focus and optimum results. In 2022, my MCN was building a career that would give me creative expression and a high income. List out possible solutions to your crucial needs Step 2: If the solutions you already know don't seem good enough, research more. In my case, baking was the only solution I knew. I had to talk to someone who introduced me to skills like Google quality rating and content writing. Identify your Low-Hanging Solution (LHS) Step 3: Your LHS is the solution with the highest probability of meeting your needs using your existing resources (skill, money, people, energy, time). This involves choosing a solution not just based on its chance for success but also considering whether you have the resources to implement it. In my case, I went with content writing because it gave me the highest chance of creative expression and high income. For resources, I already had a basic knowledge in writing and I could commit the time needed to advance the skill. Note: If none of your listed solutions are low-hanging, be open-minded and do more research to discover better ideas. Define your goal Step 4: Match your most crucial need with your lowest-hanging solution to define your goal. Simply put: MCN + LHS = Goal My goal was—to gain creative expression and increase my income through content writing. 2. Assign Specific Timeframe The benefit of having a new year goal is that it gives you a natural start and end time. But that doesn't mean it must start on January 1st and run through the year. It's ok to know what you want to do and not be ready to start it till later in the year. Only ensure you set a specific start and end date to keep you accountable. When setting your end time, ensure it's neither overstretched nor too short and leave a margin of error. How do you leave a margin of error? Set a deadline earlier than the year's end to accommodate unforeseen situations. This increases your chance of achieving your goal before the year ends. For example, I set my 2024 goals to end in the 3rd quarter of the year. This way, I can handle spillovers in the last quarter. In the absence of any spillover, evaluate your goals and plan for the next year. 3. Set Specific Outcome The truth is that setting a specific outcome can be intimidating because it leaves you no choice but to succeed. But the flexibility of vague goals makes it hard to stay committed and focused, which increases the chance of failure. So, which do you choose? If you go with specificity, keep reading to learn how to set it. Renowned entrepreneur , recommends expressing clear outcomes in absolute numbers devoid of signs like - +>/ Vusi Thembekwayo This means you should be wary of expressions that make your goal appear vague and uncertain. Such expressions often appear as phrases like: Between X and Y I'm trying to I'd like to at least more Or Here are some examples of how to take your goals from vague and uncertain to specific and confident. ❌ I'd like to improve my presentation skills. ✅ I'll enhance my presentation skills by completing a public speaking course within the next two months. ❌ I plan to learn between 2 to 5 languages this year. ✅ I'll learn Spanish and Arabic by taking online language lessons 3x a week. ❌ Maintain at least a 90% customer satisfaction rate. ✅ Achieve a customer satisfaction rate of 90% by addressing feedback and implementing improvements. ❌ I want to visit more countries. ✅ I'll visit three new countries every two months starting in April to broaden my cultural experiences. Still need a reason to set specific outcomes? According to Vusi, it enables you to give a straightforward yes or no answer when evaluating the success of your goal. This makes it easy to know the next step to take. More about it in element 9 (goal evaluation). 4. Create an Action Plan If you know your destination but don't know the route that leads there, it’s certain you'll never get there. Your goal is the destination. Your action plan is the route to achieving it. To create an action plan, identify the elements you need to achieve the goal. These elements include resources, habits, milestones, contingency plans, and scheduling. If, for instance, you have the goal of reading 15 books in various genres over the next six months, an action plan example for you will be: A list of book names, highlighter, and jotter. Resources: Process: Find book recommendations on Amazon Buy them Read them Read 20 pages daily before going to bed, except on Sunday. Habit: Complete one book every month. Milestone: If I miss a day, I'll make it up the next day by reading 40 pages. Contingency plan: Block 1 hour before bedtime on my calendar. Scheduling: Note: Before creating an action plan, conduct adequate research to know what's achievable. For instance, in the example above, it's essential to check the total pages of your book to determine the daily reading target required to meet your deadline. 5. Make it Ambitious An ambitious goal stretches you beyond your limit and makes you feel more fulfilled. For instance, you usually save $50k annually, but this year, you decided to save $100k by streamlining your expenses. But, be careful not to mistake overambitious for ambitious goals. Over-ambitious goals are impossible to achieve. E.g. Aiming to become the CEO of a Fortune 500 company within one year of starting your career. OK. This clearly looks over-ambitious, but sometimes they're not that easy to recognize. This is why you need a framework to determine the ambitiousness of your goal. The framework requires you to answer these three questions: Do you need to step out of your comfort zone or establish limits to achieve it? Yes - ambitious No - under-ambitious Despite difficulties, do you have a clear roadmap for achieving it? Yes - ambitious No - overambitious Is the reward worth the risk? Yes - ambitious No - overambitious Kindly note that the effectiveness of this framework is reliant on how well you did your research and the sincerity of your answers. 6. Develop an Adaptable Mindset After setting my 2024 career goal, I discussed it with an expert who found faults in my process and offered suggestions. It was a hard pill to swallow initially because I believed in my process. But after two days of thinking through it, I became convinced that adhering to her advice was the best decision. This underscores the importance of approaching your goal with an adaptable mindset. Certain things could distort the direction of your goal, such as: Expert advice, like in my case Unforeseen situations during implementation. Government regulations Environmental factors like the COVID pandemic It may be difficult to change your original idea if you have an affinity for it, invested efforts/resources in it, or the process of changing it is challenging. To make a rational decision, Don't settle for the first choice that comes to your mind. Instead, ponder on it. Weigh the risk of sticking with your existing plan and the advantages of adjusting it. If it's advice, like in my case, consider the person's experience and understand the reason for their suggestion. Get the opinion of a third party. Check in with your intuition. Does adaptability negate the importance of setting a specific timeframe and outcome for your goal? No! They complement each other. While the latter helps you stay committed and focused, the former keeps you attuned to reality. 7. Define Your "Why" Two months into my content writing career, I felt trapped in muddy ground. I hadn’t made a dime. I was feeling intimidated by other people's work. My family thought I was wasting my time. But I remained steadfast, and this is because I had a “why,” which is this: Writing gives me the joy of seeing my ideas breathe life. And it has the highest chance of giving me the money I need to live the life I want and make a social impact. This thought kept me motivated. This is the power of having a “why.” It motivates you to keep going, especially when the journey gets tough. Like kinetic energy is required to set an object in motion and sustain it, your "why" gives you the intrinsic energy to kick start your goal and keep the momentum. What is your "why"? It's the strongest benefit of achieving your goal. Emphasis on the word "strongest" because the stronger it is, the more steadfast you’ll be. If you choose your goal based on need, it'll be easy for you to define your "why.” When you do, keep it handy (in your subconscious mind or writing) so that you can easily refer to it when it's needed. 8. Keep it in Sight Out of sight is out of mind, they say. If you can't see your goals, there's a high chance of forgetting them. This goes beyond writing them down but embracing systems that make it easy to remember them. Tools for Keeping Your Goals in Sight: Design vision boards and hang them on prominent walls in your office or bedroom to provide motivation and help you maintain focus. Block dates in your calendar to remember what you plan to do at certain times. Establish review rituals, e.g., dedicate a day of the week or month to evaluate your progress. Write on sticky notes and place them in prominent places, such as your workspace or mirror, to serve as reminders. Share with others like friends, family, or colleagues who can provide support and hold you accountable. Use goal-tracking apps like Goals Wizard, Hatch, and Goalify for easy access, keeping track of your progress, setting reminders, and enriching your evaluation with analytics insight. 9. Evaluate Your Goal In the last quarter of the year, ask yourself this question: Did I achieve my goal? If you set a precise outcome, you should be able to answer… Yes or No If your answer is No. Don't feel bad, especially if you did your best. Rather, identify the milestones you achieved and celebrate them. And go further to assess why you couldn't hit the intended outcome. Did you choose the wrong goal? Was it overambitious? Was your action plan flawed? Were you not adaptive? Were you lazy? Did you procrastinate? Did you lose motivation? Did you forget it? I understand that it could involve a more nuanced problem. The point is for you to probe your process thoroughly to uncover mistakes and draw insights to set better goals for the next year. But if you achieved your goal, bravo 👏👏 So… What Next? There are three things you can do: Set another goal: If, after reaching this new height, you feel unsatisfied and believe there's a better chance of achieving more if you keep the momentum, set another goal. Just so you know, this is why I should have set a goal for 2023. Focus on another bucket: You can pick another bucket of your life that needs some love and set a goal for it. Don't set a goal at all: I understand that continuous goal setting can feel like an unending rat race. It's the reason the start of this article encourages you not to set goals based on necessity but on needs. So, if you feel satisfied with your current state in life, there is no pressure. Live unbound 🕊️ And know that you can always set a goal within the year if you find a need to do so.

From Failed to Achieved: 9 Proven Elements For Setting New Year Goals

After years of supporting Mozilla financially and spreading messages with my peers, I returned as an active contributor. I defined a 4-week sprint for myself that started on November 22th and finished on my birthday, December 22th. My goals were: 1: To contribute to the localization every day, including weekends. In the beginning, it was hard for me to build a habit of using the little free time I had to contribute to the L1on instead of watching a short documentary or something funnier. After the third day, I nailed it. It helped that I added a reminder for myself in my work calendar and that my current employer supported me with this mission. 2: To ensure that Thunderbird is 100% localized in Bulgarian (it was around 67% at that mark) Now it's a 100%. I could also contribute with a few strings to the Turkish and Czech translations. The next step is to ensure that the translated content makes sense to the end consumer. That's the goal of my next sprint. 3: To ensure that Firefox Focus for Android and IoS and the Facebook Container will also reach 100% localization. It's 100% now. I don't know how many people use it in Bulgaria, but if they want to, it's translated entirely. That was a concise task, thanks to the efforts of the other translators in the past. I translated the delta between January and November 4: To reconnect with the local and international community to contribute more to the project. This is where I need to invest more. I was active in the past, traveling and talking about Mozilla at local and international events and actively building the Bulgarian and Balkan communities. After that, I gave up for reasons I don't want to comment on now, partially because of my personality and ego. I've lost a lot of great friends and peers from Mozilla. I am trying to reconnect with them now. 5: To find another way to help Thunderbird apart from localization and donations. I am now active in the Thunderbird support forum and validating bugs under Linux in Bugzilla for Beta and Daily versions. I would invest some time there as well into my next sprint. Be creative, be brave, and contribute to the open source as you can with your time, knowledge, resources, or good intentions. Also published . here

Everything I Achieved With My Mozilla Contribution Sprint 

I remember when I started working as a DevOps Engineer. For me, ” best practices” was the most sacred phrase. I searched for best practices not only when I wanted to know how to do something. I searched for best practices to get inspiration for what I should do. In hindsight, big mistake. It happened because I didn’t have a clear goal. When you don’t have a clear goal, you’ll try to see what the experts do and copy. You’ll end up embracing their best practices without knowing why they did them. I’ll share with you here why “One-Click Environment” is the ultimate goal. DevOps As a bonus, I’ll share a framework to achieve it fast.‍ The ultimate DevOps goal - One-Click Environment Looking back, the goal you should set is simple: Create an environment in “One-Click”. I’m saving you years of confusion that I had to go through. Why is this the ultimate goal, you ask? Three reasons: It makes it easy to test changes It makes it clear what should be automated It improves the system’s recoverability What do I mean by “Environment”? A ‘clean’ and working version of the entire system. The most popular environment is the Production environment, which serves the clients. There are static testing environments, such as ‘dev,’ ‘staging,’ ‘uat,’ and more. There are also ephemeral testing environments, such as Pull-Request-Environments, Git-Branch-Environments, Developer-Environment, etc. Why does it make it easy to test changes? The company with the fastest development speed I ever saw was a supply-chain startup. They deployed high-quality code to production a lot, every day, with no downtime. How? We deployed ephemeral environments for every pull request. The environments were identical to production. After running automated tests, they terminated. We used Pulumi (Typescript) on AWS and documented how to change the environment. This way developers took part in contributing to the environments. Programming is not Mathematics Edsger Dijkstra was one of the pioneers of computer science. He envisioned programming as a mathematical discipline. He thought programmers would mathematically prove their code works. When was the last time you wrote a mathematical proof for your code? Probably never. Programming is Science Instead, you took the scientific approach: You ran some tests to see if your code worked as expected. And how do you run tests? On an environment, of course! The easier it is to create an environment, the faster you’ll be able to run your tests. The ‘cleaner’ your environment is, the more trust you’ll have in the test results. Why does it make it clear what should be automated? Another startup we work with has 15 Kubernetes clusters with 2,200 nodes on GCP - All created manually! Developers asked for support many times a day; "Create this DB", "I need a new NodePool", etc. With every new request, we had to ask - "Should this be automated?" We ended up automating everything required for every new environment. How? We created a Pulumi Typescript codebase and imported the entire infrastructure into it. On top of that, developers felt comfortable contributing to it, as it was a Typescript project. The “should this be automated?” mental struggle I asked myself at least 1,836 times, “Should this be automated?” on lots of things. Sometimes, I decided to automate nonsense. If you want to understand if it's worth automating, ask yourself two questions: Is it going to happen again? Will automating it take less time than doing it manually? Don’t rely on your memory You’ll need to remember every ad-hoc modification you did to an environment. If a change isn’t done as part of the “One-Click Environment” automation, you’ll have to rely on your memory and do it again. Imagine a developer creating a new environment, running its tests, only to find out the DB isn't there. Someone created it manually in the other environments. Bottom line, how do “One-Click Environments” guide automation? Next time, you ask yourself, “Should this be automated?” If it’s part of what a new and clean environment needs: Easy YES If not: A strong maybe. By the way, maybe you want to automate it, but you’re drowning in so many requests that it’s impossible. In that case, send your management . this article It's about calculating how much DevOps capacity your company needs. Maybe it’ll help you understand how much DevOps capacity you need. Why does it improve the system’s recoverability? An IoT startup we worked with had 5 clients, with an per client. AWS account It was supposed to be the same system per client, but it wasn’t! Each account’s production was created with Terraform (Terragrunt) and Ansible. But, there were so many ad-hoc changes it took about 2-4 days to create a full environment from scratch. We decided to use Jenkins and created a pipeline that automated everything. When we finished, you could deploy a full environment from scratch in 50-minutes. (after some parallelization, it went down to 20-minutes) You should have heard the sigh of relief the team had when the demo finished — ”No more fear of not recovering from a production incident!”. Create a new environment and route traffic to it Think about your production — what would it take to create it from scratch? If your answer is “I don’t know,” it’s most likely because some things were created manually or ad hoc. If your answer is “one-click,” we are hiring. Please click this link asap: . Senior DevOps Engineer Position at MeteorOps Imagine a production failure that requires creating a new environment and finding out you can't. Some things were modified manually. Or, as the poet Capone-N-Noreaga wrote - “Oh no, oh no, oh no, no, no, no no.” When you have “One-Click Environments”, you can deploy a new environment fast and use it instead of the old one.‍ It also improves high-availability When you can create an environment from scratch, things become easier: creating a multi-region setup creating an active-active setup creating an active-passive setup You can deploy many environments in different locations and route traffic to an environment based on your needs. TPCS - A useful framework to achieve a One-Click Environment Tools, Pipeline, Contributions, State ‍ It all boils down to 4 things you should be doing: You should choose the tools to manage data, provision, deploy, configure, secure, test, monitor, and automate Then, you should build the pipeline that uses the tools to create the environment You should also make it easy to contribute to the pipeline that creates an environment And you should continuously reconcile the state of the system to its desired state Principles for Choosing Tools Choose tools that answer the below requirements: More native integrations per tool, less glue required — Makes building the pipeline easier (e.g., AWS) Popular tools with active communities make documentation easier (e.g., ). Tools with a declarative syntax are self-documenting and save you some hustle (e.g., Terraform) Kubernetes Tools with native state management will save you time reconciling the state of the system (e.g., Terraform) Tools with extensive APIs will make it easier to give the developers more ownership (e.g., Pulumi) Guidelines to build the pipeline This is where you’re flexibility is your number one priority: Choose a flexible automation platform (e.g., Jenkins) Create one pipeline and make it possible to pass parameters to it Manage the pipeline’s code straight in Git Bonus: Parallelize its stages to speed it up Make it easy to contribute Describe the desired state of the system in a Git repository and make it easy for developers to contribute Make it clear where to find infra-related stuff, deployment-related stuff, etc Write good documentation ‍ Good documentation answers these questions: How to provision a resource? How to deploy a workload? How to configure the system? How to secure the system? How to test changes? How to monitor a workload? How to automate a process? How to manage the data’s lifecycle? (e.g., Run DB migrations) Reconcile the state of the system This step is made easier if you chose idempotent tools with state management, and made harder if you didn’t! (choose idempotent tools) The guiding principle here is simple. If you run the pipeline twice with the same parameters, nothing should happen the second time. That’s it; you’re good to go! But first… A Recap! Without a useful goal, you might embrace “best practices” that aren’t good for you. The most useful goal for a DevOps Engineer is being able to create an environment with “One-Click”. Because it makes it easy to test changes, And It makes it clear what should be automated, And It makes it easy to recover the system. A framework that helps achieve this goal is TPCS: Tools, Pipelines, Contributions, State ‍ Whenever I think about going forward without a goal, I’m reminded of this gem from ‘Alice in Wonderland’: “Alice: Would you tell me, please, which way I ought to go from here? The Cheshire Cat: That depends a good deal on where you want to get to. Alice: I don't much care where. The Cheshire Cat: Then it doesn't much matter which way you go. Alice: ...So long as I get somewhere. The Cheshire Cat: Oh, you're sure to do that, if only you walk long enough.” ― Lewis Carroll, Alice in Wonderland‍ Also published . here

One-Click Environments: The Ultimate DevOps Goal and How to Achieve It with TPCS Framework

You can jump to any section of y The National Cislunar Science & Technology Strateg here. Near-term Cislunar trends Soon, the United States will achieve a key milestone with the NASA Artemis program, which will send the first human-rated spacecraft to Cislunar space since 1972. Robotic exploration of Cislunar space is also growing more robust, with many orbital cubesats and the first U.S. robotic Lunar landers since the Apollo era joining a host of other global probes and missions[3] operating in Cislunar space. This increase in activity spans a range of applications for science (including Lunar mapping and surveying from orbit, surface geophysical investigations, and radio astronomy observations), in-situ resource utilization (ISRU), and other exploration and development activities. To support these planned and future Cislunar activities, basic infrastructure is needed, including communications, PNT, transportation, and radio frequency spectrum management. Near-term Cislunar enabling capabilities will need to be transparent, open, interoperable,[4] and scalable to advance a Cislunar ecosystem that is sustainable for years to come. Further, these activities must be supported by a diverse science and technology workforce equipped to tackle challenges associated with this still largely unexplored domain. [3] As one example, NASA is working with several U.S. companies to deliver science and technology to the Lunar surface through the Commercial Lunar Payload Services (CLPS) initiative. Under Artemis, commercial deliveries beginning in 2022 will perform science experiments, test technologies, and demonstrate capabilities to help NASA explore the Moon and prepare for human missions. [4] For information systems, interoperability is the capability of different systems or networks to communicate and exchange information with each other. This content was initially published on in November 2022. whitehouse.gov This content has been broken-up into bite sized sections, each with unique headlines and AI-generated lead images, in line with the provisions of Creative Commons 3.0.

NASA's Artemis Program: U.S. Set to Achieve Historic Milestone in Cislunar Space

All Images Created by The Bing Image Creator. Binary Number Representation in LLMs An Original Research Idea Binary representations can enable more efficient storage and computations compared to floating point vectors in certain cases, perhaps even the general case. Since binary vectors only need to store 0s and 1s, they require less memory and allow faster processing of certain operations like calculations. Hamming distance This could be advantageous for very large vector datasets in training, testing, deploying, and production. The number one problem with right now is the huge amount of money required to create them. LLMs If this approach is successful, that could become a thing of the past. However, converting to binary comes with some potential downsides as well. Information loss is possible when reducing a dense, high-precision floating point vector to a sparse binary representation. The binary encoding schemes would need to be carefully designed to preserve as much semantic meaning as possible. More complex vectors like those from large transformer models may be harder to binarize without degradation. For similarity search and retrieval tasks, binary representations can work well when the models are trained to produce good-quality binary codes. The key is optimizing the model to minimize information loss so that semantically similar items still map to similar binary vectors. This is an active area of research. How would such a System Work? One method is to train autoencoders to compress the dense floating point vectors from the LLM into more compact binary codes, preferably in hardware. The autoencoder's encoder portion would learn to map the high-dimensional vectors to low-dimensional binary codes, while the decoder reconstructs the original vector. The model is optimized to minimize reconstruction error, so the binary codes retain as much information as possible. Once trained, the encoder can be used to generate binary codes for new vectors. The key hyperparameters when training the autoencoder include the code size (number of bits in the binary code) and the loss function. Smaller codes are more efficient but risk higher information loss. The loss function, such as mean squared error or cosine similarity, determines what aspects of the vector are preserved. loss tends to improve semantic meaning. Cosine similarity Another approach is using hashing techniques like (LSH) to binarize vectors. locality-sensitive hashing LSH uses randomized projections to map similar vectors to the same "buckets" with high probability. The hash functions can be chosen so that the hamming distance between binary codes reflects the cosine similarity of the original vectors. At query time, binary codes for the query and database vectors can be compared via hamming distance calculations instead of cosine similarity for faster retrieval. The candidates can then be re-ranked using precise cosine scores. Training the binarization model directly on the LLM's vectors can help tailor the encoding specifically to that vector space. However, transfer learning may also work well, especially for large models like GPT-3 where training an autoencoder directly is expensive. There are also hybrid approaches possible where certain vector components are binarized while others remain as floats. For example, some semantic information could be encoded in binary while precision for positional information like word order is retained in floats. The optimal encoding scheme likely depends on the specific LLM, use case, and size of the dataset. Binary representations show promise for improving large-scale vector search and retrieval, but more research is still needed into lossy vector compression techniques for complex language model embeddings. Careful tuning and evaluation of information loss vs efficiency gains are necessary to strike the right balance. Optimizing Binary-Encoded LLMs What I am thinking of is something along these lines: - Design transformer architecture variants specialized for binary computations like binary self-attention. Could provide 10-100x speedups. Binary Transformer - Automate model design using NAS focused on binary building blocks for efficiency. Binary Neural Architecture Search - Train binary latent variable models like VAEs to compress and generate text. Binary Generative Modeling - Apply RL to learn binary policies for tasks like text summarization. Binary Reinforcement Learning - Compile models to binary circuits for customized hardware acceleration. Binary Differentiable Programming - Optimize key algorithms like PageRank and graph convolutions for binary graph embeddings. Binary Graph Algorithms - Use adversarial training to improve binary vector robustness. Binary Adversarial Regularization - Remove redundant binary weights to exploit model sparsity. Binary Pruning - Distill knowledge from larger models into quantized binary models. Binary Quantization Distillation - Apply information theory concepts like Huffman coding to activation compression. Binary Coding Theory - Binary embeddings Representing word and token embeddings in compact binary vectors instead of floats could significantly reduce memory footprint and speed up similarity calculations. (This one is so obvious!) - Using low-bit binary values for neural network activations can compress model size and enable faster integer math. Binary activations - Specialized algorithms to multiply binary matrices can optimize key computational bottlenecks. Binary matrix multiplication - Locality-sensitive hashing embeds similar items in binary codes for fast approximate nearest neighbor search. Binary hashing - Floats can be quantized to low-bit integers like 8-bit with minimal accuracy loss for 4x compression. Low-bit quantization - Skipping zero values in sparse binary vectors can avoid wasted computations. Exploit sparsity - Binary computations like matrix multiplies can be efficiently parallelized across multiple devices. Parallelize operations - Storing parameters in binary format reduces model size. Compact parameters - Dedicated hardware optimized for binary operations can provide speedups, even to 10,000x (rough estimate comparing floating point arithmetic to bit manipulation, especially with specialized hardware and dedicated vector processing or even matrix processing. Custom binary hardware Performance Speedup While theoretical peak speedups of 1000x or more may seem possible on paper, several practical engineering challenges would limit realized gains. For training, communication bottlenecks for gradient synchronization during distributed parallel training can constrain speedups. While binary representations reduce communication bandwidth needs, the core calculation throughput will be dependent on parallelizing across multiple devices. Data parallelism helps, but model parallel approaches may be needed to fully leverage thousands of devices in parallel. This introduces additional software complexity. The degree of binarization possible is also limited by accuracy considerations. Certain components like attention mechanisms are harder to fully binarize without impacting model quality. . A hybrid approach likely works better, combining binary and floating point operations. Finding the right balance for each model is key Custom hardware like ASICs can accelerate binary matrix multiplications and other basic ops considerably. But they sacrifice programmability and make deploying improved model architectures harder. FPGAs may provide more flexibility but lower peak throughput. Production deployment at scale introduces additional systems engineering challenges around low-latency parallel inference. Careful optimization of data pipelines, batching strategies, and hardware provisioning is essential to maximize throughput without latency spikes. So while binary optimization can provide order-of-magnitude improvements, the practical limits depend on model architecture, hardware mix, use case tradeoffs between latency and throughput, and engineering effort. A nuanced assessment is required to determine realistic speedup potential for a given large language model workflow. But this remains an exciting and promising area for continued research! Transformeromorphic Computing Yes, the term does not exist. I coined it just now. The inspiration comes from Neuromorphic Computing. Neuromorphic computing aims to mimic biological and could potentially provide benefits like massive parallelism, low power consumption, and optimized performance for neural network-style computationsc. neural networks This makes it an intriguing approach to explore for accelerating large language models that are based on deep neural networks. Some potential ways neuromorphic hardware could help optimize large language models: Custom spiking neural network architectures that can efficiently run transformer-based models while taking advantage of neuromorphic parallelism and event-driven processing. Novel training methods and algorithms tailored for sparse, asynchronous neuromorphic hardware. This could allow for faster and more efficient training. Specialized neuromorphic chips designed to accelerate matrix multiplication, attention mechanisms, and other core operations used in large language models. On-chip learning techniques to update parameters and enable continuous adaptation on deployed neuromorphic systems. Low-power neuromorphic accelerators for efficient inference of large language models. Neuromorphic computing remains an exciting area of research for next-generation AI acceleration. : Parallel Processing Neuromorphic systems are inherently parallel, much like the brain. This means they can process many operations simultaneously, unlike traditional CPUs that are largely sequential. This could lead to massive speed-ups in tasks that are parallelizable, such as processing different parts of a language model or handling multiple requests at once. : Energy Efficiency Neuromorphic chips can be much more energy-efficient than traditional hardware, as they only consume power when needed, similar to neurons that only fire when activated. This efficiency could allow for larger models to be run on less power, reducing costs and enabling more sustainable operations. : Low-Latency Processing Due to their design, neuromorphic systems can have very low latency, which is beneficial for real-time language processing tasks. This could improve the responsiveness of language models in applications like translation, voice assistants, and interactive chatbots. : Sparsity Exploitation Language models often deal with sparse data, where most of the elements are zeros. Neuromorphic hardware can exploit this sparsity to skip unnecessary computations, leading to speed-ups. This is because they can be designed to only process and transmit non-zero data, much like how neurons only fire when a certain threshold is reached. : On-chip Learning Neuromorphic systems can potentially support on-chip learning, which means they can adapt and optimize in real-time without the need for extensive data transfer between the processor and memory. This could reduce the time and resources needed to update language models. : Analog Computation If the neuromorphic hardware uses analog computation for certain tasks, it could perform operations faster and with less energy than digital computation, especially for operations that are not binary by nature. Estimating the speed-ups is challenging without specific hardware and model details, but considering the parallel processing capabilities alone, one could expect orders of magnitude in performance improvement for certain tasks. For example, if a task that previously took hours on a traditional CPU could be parallelized across thousands of neuromorphic cores, it might be completed in minutes or even seconds. The implications are staggering! GPU and TPU Usage Using GPUs and TPUs  for processing binary data instead of floating-point data can indeed be a game changer in certain contexts. GPUs and TPUs are highly parallel processors that are well-suited for handling large blocks of data and performing the same operation on multiple data points simultaneously. Here's how using GPUs/TPUs with binary data could be particularly advantageous: : Binary operations are simpler and faster than floating-point operations. GPUs, with their thousands of cores, could perform binary operations at much higher throughput, potentially leading to significant speed-ups in processing times. Increased Throughput : Binary data requires less precision compared to floating-point data. This means that the GPU's memory bandwidth and storage could be used more efficiently, as binary data takes up less space. This efficiency could allow for larger datasets to be processed in parallel, or for more data to be stored in the faster levels of the memory hierarchy (like cache), leading to performance improvements. Reduced Precision Requirements : Binary operations consume less power than floating-point operations. When running on GPUs, this could lead to more energy-efficient processing, which is crucial for large-scale computations and for environments where power consumption is a concern. Energy Efficiency : The complexity of algorithms could be reduced when dealing with binary data. This simplification could lead to easier optimization and implementation of algorithms on GPUs, potentially reducing development time and errors. Simplified Computation : If all data were binary, there could be a push for GPUs or other processors to be optimized specifically for binary computations. This could lead to specialized hardware that is even more efficient at handling binary data than current GPUs. Potential for Specialized Hardware : Binary data could lead to more uniform and predictable computation times, which is ideal for the highly parallel architecture of GPUs/TPUs. This could improve the efficiency of GPU utilization and reduce the occurrence of some cores being idle while others are still working. Improved Parallelism In the context of large language models and other AI applications, the use of binary data could be more nuanced. Techniques like quantization already aim to reduce the precision of data to speed up computation and reduce model size, but this is a trade-off between performance and accuracy. If a language model could be effectively run using binary data without a significant loss in accuracy, then using GPUs could indeed be a game changer in terms of speed and efficiency. Conclusion In conclusion, the exploration of neuromorphic hardware and GPUs in the context of optimizing large language models presents a fascinating glimpse into the future of computational efficiency and performance. GPUs/TPUs, which are already a cornerstone of high-performance computing, particularly in the realm of machine learning and data processing, could see a transformative leap in their utility if the data they processed were binary. The simplicity and speed of binary operations, combined with the massive parallel processing capabilities of these devices, could lead to unprecedented throughput and energy efficiency. This would be a game changer, especially in an era where the volume of data is continuously growing, and the need for faster, more efficient processing is ever-present. However, it's crucial to recognize that the shift to binary data would involve trade-offs, particularly in the precision and granularity of information that floating-point numbers provide. The real-world applicability of such a shift would depend on the specific requirements of each application and the extent to which accuracy can be maintained. The potential for neuromorphic hardware and optimized GPU/TPU usage in processing binary data is a testament to the innovative spirit that drives the software engineering industry. It underscores the importance of quality assurance, time management, and a user-centric approach in developing technologies that can meet the demands of the future. As we continue to push the boundaries of what's possible, the role of software engineers will be integral in shaping a world where these advanced computational strategies become a reality, enhancing the capabilities of large language models and the myriad of applications they power. All images Created by the Bing Image Creator.

How to Achieve 1000x LLM Speed for Efficient and Cost-Effective Training, Testing, and Deployment

দক্ষ এবং খরচ-কার্যকর প্রশিক্ষণ, পরীক্ষা এবং স্থাপনার জন্য কিভাবে 1000x LLM গতি অর্জন করা যায়

So erreichen Sie die 1000-fache LLM-Geschwindigkeit für effizientes und kostengünstiges Training, Tests und Bereitstellung

Cómo lograr una velocidad de LLM 1000x para una capacitación, pruebas e implementación eficientes y rentables

Comment atteindre une vitesse LLM 1 000 fois supérieure pour une formation, des tests et un déploiement efficaces et rentables

कुशल और लागत प्रभावी प्रशिक्षण, परीक्षण और तैनाती के लिए 1000x एलएलएम गति कैसे प्राप्त करें

効率的かつコスト効率の高いトレーニング、テスト、導入のために 1000 倍の LLM 速度を達成する方法

효율적이고 비용 효율적인 교육, 테스트 및 배포를 위해 1000x LLM 속도를 달성하는 방법

Como alcançar velocidade LLM 1000x para treinamento, testes e implantação eficientes e econômicos

Как добиться 1000-кратного увеличения скорости LLM для эффективного и экономичного обучения, тестирования и развертывания

Verimli ve Uygun Maliyetli Eğitim, Test ve Dağıtım için 1000 Kat LLM Hızına Nasıl Ulaşılır?

Cách đạt được tốc độ LLM 1000 lần để đào tạo, thử nghiệm và triển khai hiệu quả và tiết kiệm chi phí

如何实现 1000 倍的 LLM 速度以实现高效且经济高效的培训、测试和部署

By its very nature, crypto attracts all walks of life. Participation in the digital economy , and enjoys representation . This makes sense when considering the motivations that lead people to take their value outside of traditional markets. The ongoing uncertainty and lack of accountability for TradFi’s bad actors continue to be sources of anxiety for retail customers. Bitcoin’s genesis block explicit mention of the failures that transpired following the 2008 global financial crisis. From its onset, this skepticism and hope for a system that puts people ahead of privileged insiders was baked into crypto’s source code. registers across age demographics throughout the political spectrum best captures this mood through However, the response from early enthusiasts within the crypto space set a tone that eschewed most aspects of traditional finance, including the scant protections that existed. Unfortunately, such “ ” thinking continues to draw negative attention from regulators while re-establishing an individualist pecking order that remains an albatross for traditional finance. While Satoshi Yakamoto’s preaches “trustless” transactions on a peer-to-peer network, it does not suggest participants should silo themselves in opposition to the broader community (see the BTC genesis block). Rather, allowing a mindset predicated on self-interest to take root risks polluting the inherent potential of crypto’s innovative technology. Like many instances in the real world, emphasis on the individual frequently holds back progress on a systemic level, even in instances where those same individuals stand to gain. A great example, at least in the United States, is education. Wild West white paper In many U.S. communities, within its zone. Along with funding for libraries and parks, these amenities aim to serve everyone and function as pillars of connection within a community. Too often, residents without children (and run-of-the-mill curmudgeons) will raise the argument that, since they experience no direct benefit from these tax-funded institutions, they shouldn’t have to contribute. It’s a somewhat sound argument if you squint because who needs the fire department when your house isn’t actively burning, right? What this umbrage fails to consider is how these investments are injected into the community through a variety of means. a portion of property taxes are allocated to schools back Obstinance to this process from well-off homeowners is inherently a net drain on the surrounding area despite its obvious benefits. Not only are homes situated closer to schools , but desirable (read: well-funded) districts that gallop prices well above market rates. And this is all before addressing the societal benefits that are created, which can have lasting effects on a macro level. typically valued higher can incite bidding wars From a shrewd (read: economic) perspective, schools are job creators and serve as integral nodes that support a community’s potential to deploy a workforce across any number of industries. When properly resourced, schools reciprocate that value by , and prepares subsequent generations for adult life. If your blood pressure is rising, pause and think of a teacher who helped you become the person you are today. Ask yourself: aren’t children more likely to succeed when they’re well-educated supported? , offering a variety of programming that alleviates working parents promotes social engagement for elderly citizens and Now imagine a society where cantankerous taxpayers get their way and retain their wealth without these contributions to the public good. Like us all, they eventually grow old. Nestled in their high-value homes after decades of denying resources to the surrounding blocks, they start to feel the walls closing in. Sure, they’re comfortable. But have trickled up, making it difficult to find local home care that meets their unique needs. This is compounded by a rising lack of agency. At first, just driving at night is tricky. declining graduation rates But soon, the car is safest in park. Homebound, they start to look around their neighborhood for outlets. Available resources that were viewed as plentiful, back when contributing to their longevity felt superfluous, have dried up. And where it was once unconscionable why anyone would go see a high school matinee of , such a social experience suddenly feels like a luxury. Bye Bye Birdie From this commonly accessible vantage point, a two-pronged approach to reorienting the crypto ecosystem emerges: nurture a true sense of community and find points of commonality that rally solidarity among participants. Like the school example above, such a holistic approach could reimagine how networks interact and benefit in the digital economy by abandoning untoward self-interest. If business-as-usual ideology results in regular and in traditional finance (and, let’s be honest, ), why replicate such norms? Why build an analogous system to the one whose colossal failure sparked Satoshi’s monumental breakthrough? calamities fraud crypto too Rather than waiting until the house is on fire or we’re stuck longing for amateur theater, leaders need to think proactively about building solutions that account for crypto’s ultimate dream. Despite effective altruism proving itself to be little more than robber baron alimony for the digital age, . Carnegie at least left libraries behind From equitably sharing the spoils to ceasing black hat trading techniques to rooting out opportunists who see the world as a mark, a new vision must rise from its ashes. This is both a big tent technological vision and a new covenant of values. To find firm global footing, crypto must retreat from gluttonous visions of wealth-at-all-costs if it hopes to survive . compounding regulatory and consumer culls To achieve global adoption, crypto companies need to start conceiving of the ecosystem as a single organism whose success requires collaborative effort and investment. Whether this demands the development of new or better networks, true progress must be measured by our collective will to achieve financial freedom. No longer a zero-sum game, this is an opportunity for the best and brightest to determine which stars are within reach. After watching how the class of Web 2.0 matured and the world they left behind, it's clear that what comes next must truly aim to shift the cultural paradigm.

To Achieve Global Adoption, Crypto Should Remember Its Roots

গ্লোবাল অ্যাডপশন অর্জন করতে, ক্রিপ্টোকে তার শিকড়গুলি মনে রাখা উচিত

Um eine weltweite Akzeptanz zu erreichen, sollte sich Krypto an seine Wurzeln erinnern

Para lograr una adopción global, las criptomonedas deben recordar sus raíces

Pour parvenir à une adoption mondiale, la crypto devrait se souvenir de ses racines

वैश्विक स्वीकृति प्राप्त करने के लिए, क्रिप्टो को अपनी जड़ें याद रखनी चाहिए

世界的な普及を達成するには、暗号通貨はそのルーツを思い出す必要があります

글로벌 채택을 달성하려면 암호화폐는 그 뿌리를 기억해야 합니다

Para alcançar a adoção global, a criptografia deve se lembrar de suas raízes

Чтобы добиться глобального принятия, криптовалюта должна помнить свои корни

Küresel Benimsenmeyi Sağlamak İçin Kripto, Köklerini Hatırlamalı

Để đạt được sự chấp nhận toàn cầu, tiền điện tử nên nhớ về nguồn gốc của nó

为了实现全球采用，加密货币应该记住它的根源

AI can automate various mundane aspects of the cryptocurrency ecosystem, such as customer support, document verification, and data analysis, thereby increasing efficiency and reducing operational costs. But, there is so much more the merger of AI and crypto could achieve. The integration of these technologies will transform the crypto landscape. Machine learning algorithms, for instance, are being used to improve security and to analyze tons of data with the goal of detecting fraudulent activities and potential security breaches within the blockchain ecosystem. Algorithms can identify suspicious activities in real-time, flagging potential fraudulent behavior and mitigating risks. This could enhance the overall trustworthiness and reliability of cryptocurrencies. Investment Analysis Trading bots, improved data analysis, and machine learning algorithms are helping traders to analyze data sets and providing insights into cryptocurrency price movements. These technologies aid traders in maximizing their profits and even providing personalized cryptocurrency and blockchain-related services, such as tailored investment recommendations, based on an individual's financial goals and risk tolerance. Crypto traders and investors of the future will increasingly employ such tools to gain insights into market trends and to make more informed decisions. This technology can be used to assess the viability of ICOs and tokens. By analyzing data about the project, team, and market sentiment, Tools can provide insights into the potential success of an ICO. Another technology is Natural Language Processing (NLP) technology, which is used already in sentiment analysis of news, social media, and other online content to gauge public sentiment around cryptocurrencies. This information can be valuable for traders and investors. Machine intelligence, however, could have a negative effect on markets. High-frequency trading algorithms powered by AI could significantly impact market liquidity and trading volume in the crypto space. These algorithms can execute trades at high speeds, affecting price movements. Enhancing Security, Efficiency, And Risk Assessment Cryptocurrencies can be susceptible to fraud and cyberattacks. Modern technologies can detect and prevent fraudulent activities by identifying suspicious transactions and wallets. Such systems can provide risk assessment and help traders and investors manage their portfolios more effectively, improving the security of cryptocurrency platforms. Machine learning algorithms can analyze user behavior to identify anomalies and potential security breaches. Blockchain Data Analysis This new technology can analyze the blockchain to extract insights and patterns, helping out with audits, tracking transactions, and gaining a deeper understanding of how cryptocurrencies are being used. AI Algorithms have proven to be highly effective in detecting fraudulent activities, ensuring the integrity of transactions, and safeguarding digital assets. The technology can identify patterns, anomalies, and potential threats that may go unnoticed by human analysts. Systems could monitor blockchain networks and swiftly detect any suspicious activities or attempted hacks, while also streamlining transaction processes by automating routine tasks such as identity verification and compliance checks. This not only reduces the likelihood of human error, but also accelerates transaction speeds. AI Can Enhance Security Measures AI can enhance security measures by identifying potential threats and vulnerabilities in blockchain networks, improve transaction speed and accuracy, minimize human error, as well as assist in regulatory compliance by monitoring transactions for potential money laundering and ensuring adherence to financial regulations in the cryptocurrency sector. There is potential for the centralization of power in AI-crypto systems, as a few entities could control a significant portion of network operations. There are ethical considerations, too, regarding the use of such technology in making financial decisions that might affect individuals' investments or even market stability. Trading algorithms can contribute to market volatility and sudden price fluctuations. At the end of the day, AI is continuously evolving, and its impact on the crypto space is still unknown. From investment analysis to security and provenance, the impact crypto and AI will have on each other could transform both industries.

AI and Crypto Could Achieve So Much More Together

USA v. Google LLC Court Filing, retrieved on January 24, 2023 is part of . You can jump to any part in this filing . This is part 10 of 44. HackerNoon’s Legal PDF Series here IV. GOOGLE’S SCHEME TO DOMINATE THE AD TECH STACK A. Google Buys Control of the Key Tools that Link Publishers and Advertisers 76. Google entered display advertising on the back of its early strength in search and search advertising. In 2000, Google launched Google Ads (then called “AdWords”), a selfservice buying tool for advertisers. At the time, advertisers could use Google Ads to purchase advertising on the webpage displaying Google search results. 77. As Google’s search engine dominance grew, it attracted large numbers of small and large businesses that considered advertising on Google’s search results page to be critical to reaching customers searching for their products or services. After amassing this pool of advertisers, Google realized it could not only sell them advertising space on Google’s search results page, but also step in as an intermediary to sell them advertising space on non-Google websites as well. Thus in 2003, Google changed the default setting on Google Ads so that businesses were automatically opted into using Google Ads to advertise on third-party websites through what became known as or “ .” Today, Google Ads has grown to represent over two million advertisers, spending about $11 billion worldwide on open web display inventory per year. Google Ads is a substantial, unique source of advertising demand and revenue for publishers. Google Display Network, GDN 78. In 2006, Google found itself without sufficient access to non-Google premium advertising inventory to meet its advertisers’ demand. Effectively integrating Google Ads with existing publisher-facing platforms would have benefited both Google Ads advertisers—by increasing their access to inventory—and Google—by increasing advertising sales, and in turn Google’s total revenues as a percentage of those sales. Instead, Google sought to maintain more control over advertising purchases made by its Google Ads’ advertisers. In particular, it limited the ability of its Google Ads’ advertisers to buy inventory from Google’s rivals. Google recognized that if it could secure access to its own pool of publisher inventory, it could control the entire transaction, end-to-end, and become the “the be-all, and end-all location for all ad serving.” To that end, Google built and launched its own publisher ad server, but the product failed to gain traction. 79. Rather than innovate and compete, Google found a shortcut. In 2007, Google announced that it would buy DoubleClick for $3.1 billion. DoubleClick offered the industryleading publisher ad server, called DoubleClick for Publishers or “DFP”, which at the time had an estimated 60% market share. DoubleClick also was developing a nascent ad exchange, called AdX. 80. The DoubleClick acquisition was a pivotal moment for Google’s display advertising technology business and its strategy to dominate the ad tech stack. The deal provided Google with direct access to website publishers (and their inventory) on DoubleClick’s publisher ad server and, for the first time, a significant presence on both the advertiser and publisher sides of the ad tech stack. Google feared that if a rival acquired DoubleClick, Google would not control all the tools that link Google’s advertisers with publisher inventory; in short, a rival could “disintermediate” Google. Disintermediation risked allowing another company to control how and where publishers sell impressions to advertisers, something Google would not tolerate because it would limit Google’s ability to generate monopoly profits. Setting the stage for what was to come, the DoubleClick acquisition provided Google the unilateral power to implement a series of anticompetitive restraints, using its dominance on both the publisher and advertiser sides of the market to inhibit competition across the entire ad tech stack. 81. The Federal Trade Commission (“FTC”) investigated Google’s proposed acquisition of DoubleClick. The FTC considered “the possibility that Google could leverage DoubleClick’s leading position in third party ad serving to its advantage in the ad intermediation market” and whether Google could “exclusively bundle AdWords [advertiser demand] with [its publisher-side platforms] AdSense and DFP.” 82. The FTC ultimately declined to challenge Google’s acquisition of DoubleClick for the reasons set out in its public closing statement. The FTC concluded that “DoubleClick does not have market power despite its high market share”—over 60% at the time—and that “firms can and do switch ad serving firms when it is in their self-interest to do so.” Based on these assumptions, the FTC believed any anticompetitive conduct by Google “would likely be defeated by customers switching to one of the other third-party ad serving products.” 83. Google’s contemporaneous business documents paint a very different picture, however. Six months after the FTC closed its investigation without taking action, one senior Google executive wrote about the importance of controlling access to publisher inventory through the DFP publisher ad server, stating “the thing we want ‘secured’ is the DFP platform adoption” because “if we have this” then Google would “have a +20% monetization advantage.” In other words, Google believed it could sustainably charge a higher price on advertising transactions than its competitors because it controlled the process and rules by which publisher inventory could be sold. 84. Google knew the emerging ad tech market better, and acknowledged in internal documents that “due to [the publisher ad server’s] position as the operating system for ad sales, switching costs are very high.” Because of this “stickiness” of publisher ad servers, Google knew it could manipulate the system in its favor—and to the detriment of competition—without fear that publishers would switch to other publisher ad servers. After the acquisition was completed, the former DoubleClick CEO explained at an internal Google strategy meeting, “My view is nothing really matters but the platform [publisher ad server]. Nothing has such high switching costs. If there’s a better network or exchange, you can just switch to it. Switching platforms is a nightmare. Takes an act of God to do it.” 85. Following the DoubleClick purchase, Google cemented its position as the dominant intermediary between advertisers and publishers through a series of additional acquisitions that eliminated potential competitors and further bolstered Google’s position in open digital advertising. For example, in 2009, Google paid $750 million to purchase AdMob, a technology system that allowed publishers of mobile apps to sell ads as well. While Google’s conduct in the distinct market for mobile app advertising is outside the scope of this Complaint, Google’s anticompetitive conduct in the mobile apps market is consistent with the conduct alleged in the market for display advertising. 86. In 2010, Google acquired Invite Media for approximately $81 million. Invite Media offered a demand side platform. Google subsumed Invite Media into a demand side platform it was developing, Display & Video 360 (at the time, known as “DoubleClick Bid Manager”). By capturing an increasingly large share of bigger, more sophisticated advertisers and advertising agencies, Display & Video 360 complemented Google Ads and expanded Google’s control over advertiser demand. 87. In 2011, Google bought AdMeld for approximately $400 million. As discussed further below, AdMeld had developed technology to provide “yield management” functionality to publishers. Yield managers like AdMeld helped publishers manage inventory and optimize revenue by comparing offers from multiple advertiser demand sources at the same time. This comparison feature made it easier for new ad exchanges and advertiser demand sources to enter the ad tech industry because it gave publishers the incentive and ability to switch between ad exchanges and advertiser demand sources in response to better prices and service. 88. The DoubleClick, Invite Media, and AdMeld acquisitions helped Google achieve dominant positions at each level of the open web ad tech stack and set the stage for Google to control and manipulate the process by which publishers sell and advertisers buy open web display inventory. Continue Reading . Here About HackerNoon Legal PDF Series: We bring you the most important technical and insightful public domain court case filings. This court case 1:23-cv-00108 retrieved on September 8, 2023, from is part of the public domain. The court-created documents are works of the federal government, and under copyright law, are automatically placed in the public domain and may be shared without legal restriction. justice.gov

Google Achieved Market Dominance by Acquiring DoubleClick, Invite Media, & AdMeld, to name a few 

Google logró el dominio del mercado al adquirir DoubleClick, Invite Media y AdMeld, por nombrar algunos

Google a atteint sa domination sur le marché en acquérant DoubleClick, Invite Media et AdMeld, pour n'en nommer que quelques-uns

Google ने कुछ नाम रखने के लिए डबलक्लिक, इनवाइट मीडिया और एडमेल्ड का अधिग्रहण करके बाज़ार प्रभुत्व हासिल किया

Google は、数例を挙げると DoubleClick、Invite Media、AdMeld を買収することで市場の優位性を達成しました

O Google alcançou domínio de mercado ao adquirir DoubleClick, Invitation Media e AdMeld, para citar alguns

Google đã đạt được sự thống trị thị trường bằng cách mua lại DoubleClick, Mời truyền thông và AdMeld, cùng một số tên tuổi khác

Google 通过收购 DoubleClick、Invite Media 和 AdMeld 等公司实现了市场主导地位

Christian Science by Mark Twain, is part of the HackerNoon Books Series. You can jump to any chapter in this book . CHAPTER II here CHAPTER II When she wrote this little biography her great life-work had already been achieved, she was become renowned; to multitudes of reverent disciples she was a sacred personage, a familiar of God, and His inspired channel of communication with the human race. Also, to them these following things were facts, and not doubted: She had written a Bible in middle age, and had published it; she had recast it, enlarged it, and published it again; she had not stopped there, but had enlarged it further, polished its phrasing, improved its form, and published it yet again. It was at last become a compact, grammatical, dignified, and workman-like body of literature. This was good training, persistent training; and in all arts it is training that brings the art to perfection. We are now confronted with one of the most teasing and baffling riddles of Mrs. Eddy's history—a riddle which may be formulated thus: How is it that a primitive literary gun which began as a hundred-yard flint-lock smooth-bore muzzle-loader, and in the course of forty years has acquired one notable improvement after another—percussion cap; fixed cartridge; rifled barrel; efficiency at half a mile how is it that such a gun, sufficiently good on an elephant hunt (Christian Science) from the beginning, and growing better and better all the time during forty years, has always collapsed back to its original flint-lock estate the moment the huntress trained it on any other creature than an elephant? Something more than a generation ago Mrs. Eddy went out with her flint-lock on the rabbit range; and this was a part of the result: “After his decease, and a severe casualty deemed fatal by skilful physicians, we discovered that the Principle of all healing and the law that governs it is God, a divine Principle, and a spiritual not material law, and regained health.”—Preface to Science and Health, first revision, 1883. N.B. Not from the book itself; from the Preface. You will notice the awkwardness of that English. If you should carry that paragraph up to the Supreme Court of the United States in order to find out for good and all whether the fatal casualty happened to the dead man—as the paragraph almost asserts—or to some person or persons not even hinted at in the paragraph, the Supreme Court would be obliged to say that the evidence established nothing with certainty except that there had been a casualty—victim not known. The context thinks it explains who the victim was, but it does nothing of the kind. It furnishes some guessing-material of a sort which enables you to infer that it was “we” that suffered the mentioned injury, but if you should carry the language to a court you would not be able to prove that it necessarily meant that. “We” are Mrs. Eddy; a funny little affectation. She replaced it later with the more dignified third person. The quoted paragraph is from Mrs. Eddy's preface to the first revision of Science and Health (1883). Sixty-four pages further along—in the body of the book (the elephant-range), she went out with that same flint-lock and got this following result. Its English is very nearly as straight and clean and competent as is the English of the latest revision of Science and Health after the gun has been improved from smooth-bore musket up to globe-sighted, long distance rifle: “Man controlled by his Maker has no physical suffering. His body is harmonious, his days are multiplying instead of diminishing, he is journeying towards Life instead of death, and bringing out the new man and crucifying the old affections, cutting them off in every material direction until he learns the utter supremacy of Spirit and yields obedience thereto.” In the latest revision of Science and Health (1902), the perfected gun furnishes the following. The English is clean, compact, dignified, almost perfect. But it is observable that it is not prominently better than it is in the above paragraph, which was a product of the primitive flint-lock: “How unreasonable is the belief that we are wearing out life and hastening to death, and at the same time we are communing with immortality? If the departed are in rapport with mortality, or matter, they are not spiritual, but must still be mortal, sinful, suffering, and dying. Then wherefore look to them—even were communication possible—for proofs of immortality and accept them as oracles?” —Edition of 1902, page 78. With the above paragraphs compare these that follow. It is Mrs. Eddy writing—after a good long twenty years of pen-practice. Compare also with the alleged Poems already quoted. The prominent characteristic of the Poems is affectation, artificiality; their makeup is a complacent and pretentious outpour of false figures and fine writing, in the sophomoric style. The same qualities and the same style will be found, unchanged, unbettered, in these following paragraphs—after a lapse of more than fifty years, and after—as aforesaid—long literary training. The italics are mine: “What plague spot or bacilli were [sic] gnawing [sic] at the heart of this metropolis... and bringing it [the heart] on bended knee? Why, it was an institute that had entered its vitals—that, among other things, taught games,” et cetera.—C.S. Journal, p. 670, article entitled “A Narrative—by Mary Baker G. Eddy.” “Parks sprang up [sic]... electric-cars run [sic] merrily through several streets, concrete sidewalks and macadamized roads dotted [sic] the place,” et cetera.—Ibid. “Shorn [sic] of its suburbs it had indeed little left to admire, save to [sic] such as fancy a skeleton above ground breathing [sic] slowly through a barren [sic] breast.”—Ibid. This is not English—I mean, grown-up English. But it is fifteen-year-old English, and has not grown a month since the same mind produced the Poems. The standard of the Poems and of the plague-spot-and-bacilli effort is exactly the same. It is most strange that the same intellect that worded the simple and self-contained and clean-cut paragraph beginning with “How unreasonable is the belief,” should in the very same lustrum discharge upon the world such a verbal chaos as the utterance concerning that plague-spot or bacilli which were gnawing at the insides of the metropolis and bringing its heart on bended knee, thus exposing to the eye the rest of the skeleton breathing slowly through a barren breast. The immense contrast between the legitimate English of Science and Health and the bastard English of Mrs. Eddy's miscellaneous work, and between the maturity of the one diction and the juvenility of the other, suggests—compels—the question, Are there two guns? It would seem so. Is there a poor, foolish, old, scattering flint-lock for rabbit, and a long-range, centre-driving, up-to-date Mauser-magazine for elephant? It looks like it. For it is observable that in Science and Health (the elephant-ground) the practice was good at the start and has remained so, and that the practice in the miscellaneous, outside, small-game field was very bad at the start and was never less bad at any later time. I wish to say that of Mrs. Eddy I am not requiring perfect English, but only good English. No one can write perfect English and keep it up through a stretch of ten chapters. It has never been done. It was approached in the “well of English undefiled”; it has been approached in Mrs. Eddy's Annex to that Book; it has been approached in several English grammars; I have even approached it myself; but none of us has made port. Now, the English of Science and Health is good. In passages to be found in Mrs. Eddy's Autobiography (on pages 53, 57, 101, and 113), and on page 6 of her squalid preface to Science and Health, first revision, she seems to me to claim the whole and sole authorship of the book. That she wrote the Autobiography, and that preface, and the Poems, and the Plague-spot-Bacilli, we are not permitted to doubt. Indeed, we know she wrote them. But the very certainty that she wrote these things compels a doubt that she wrote Science and Health. She is guilty of little awkwardnesses of expression in the Autobiography which a practiced pen would hardly allow to go uncorrected in even a hasty private letter, and could not dream of passing by uncorrected in passages intended for print. But she passes them placidly by; as placidly as if she did not suspect that they were offenses against third-class English. I think that that placidity was born of that very unawareness, so to speak. I will cite a few instances from the Autobiography. The italics are mine: “I remember reading in my childhood certain manuscripts containing Scriptural Sonnets, besides other verses and enigmas,” etc. Page 7. [On page 27.] “Many pale cripples went into the Church leaning on crutches who came out carrying them on their shoulders.” It is awkward, because at the first glance it seems to say that the cripples went in leaning on crutches which went out carrying the cripples on their shoulders. It would have cost her no trouble to put her “who” after her “cripples.” I blame her a little; I think her proof-reader should have been shot. We may let her capital C pass, but it is another awkwardness, for she is talking about a building, not about a religious society. “Marriage and Parentage” [Chapter-heading. Page 30]. You imagine that she is going to begin a talk about her marriage and finish with some account of her father and mother. And so you will be deceived. “Marriage” was right, but “Parentage” was not the best word for the rest of the record. It refers to the birth of her own child. After a certain period of time “my babe was born.” Marriage and Motherhood—Marriage and Maternity—Marriage and Product—Marriage and Dividend—either of these would have fitted the facts and made the matter clear. “Without my knowledge he was appointed a guardian.” Page 32. She is speaking of her child. She means that a guardian for her child was appointed, but that isn't what she says. “If spiritual conclusions are separated from their premises, the nexus is lost, and the argument with its rightful conclusions, becomes correspondingly obscure.” Page 34. We shall never know why she put the word “correspondingly” in there. Any fine, large word would have answered just as well: psychosuperintangibly—electroincandescently—oligarcheologically— sanchrosynchro-stereoptically—any of these would have answered, any of these would have filled the void. “His spiritual noumenon and phenomenon silenced portraiture.” Page 34. Yet she says she forgot everything she knew, when she discovered Christian Science. I realize that noumenon is a daisy; and I will not deny that I shall use it whenever I am in a company which I think I can embarrass with it; but, at the same time, I think it is out of place among friends in an autobiography. There, I think a person ought not to have anything up his sleeve. It undermines confidence. But my dissatisfaction with the quoted passage is not on account of noumenon; it is on account of the misuse of the word “silenced.” You cannot silence portraiture with a noumenon; if portraiture should make a noise, a way could be found to silence it, but even then it could not be done with a noumenon. Not even with a brick, some authorities think. “It may be that the mortal life-battle still wages,” etc. Page 35. That is clumsy. Battles do not wage, battles are waged. Mrs. Eddy has one very curious and interesting peculiarity: whenever she notices that she is chortling along without saying anything, she pulls up with a sudden “God is over us all,” or some other sounding irrelevancy, and for the moment it seems to light up the whole district; then, before you can recover from the shock, she goes flitting pleasantly and meaninglessly along again, and you hurry hopefully after her, thinking you are going to get something this time; but as soon as she has led you far enough away from her turkey lot she takes to a tree. Whenever she discovers that she is getting pretty disconnected, she couples-up with an ostentatious “But” which has nothing to do with anything that went before or is to come after, then she hitches some empties to the train-unrelated verses from the Bible, usually—and steams out of sight and leaves you wondering how she did that clever thing. For striking instances, see bottom paragraph on page 34 and the paragraph on page 35 of her Autobiography. She has a purpose—a deep and dark and artful purpose—in what she is saying in the first paragraph, and you guess what it is, but that is due to your own talent, not hers; she has made it as obscure as language could do it. The other paragraph has no meaning and no discoverable intention. It is merely one of her God-over-alls. I cannot spare room for it in this place. “I beheld with ineffable awe our great Master's marvelous skill in demanding neither obedience to hygienic laws nor,” etc. Page 41. The word is loosely chosen-skill. She probably meant judgment, intuition, penetration, or wisdom. “Naturally, my first jottings were but efforts to express in feeble diction Truth's ultimate.” Page 42. One understands what she means, but she should have been able to say what she meant—at any time before she discovered Christian Science and forgot everything she knew—and after it, too. If she had put “feeble” in front of “efforts” and then left out “in” and “diction,” she would have scored. “... its written expression increases in perfection under the guidance of the great Master.” Page 43. It is an error. Not even in those advantageous circumstances can increase be added to perfection. “Evil is not mastered by evil; it can only be overcome with Good. This brings out the nothingness of evil, and the eternal Somethingness vindicates the Divine Principle and improves the race of Adam.” Page 76. This is too extraneous for me. That is the trouble with Mrs. Eddy when she sets out to explain an over-large exhibit: the minute you think the light is bursting upon you the candle goes out and your mind begins to wander. “No one else can drain the cup which I have drunk to the dregs, as the discoverer and teacher of Christian Science” Page 47. That is saying we cannot empty an empty cup. We knew it before; and we know she meant to tell us that that particular cup is going to remain empty. That is, we think that that was the idea, but we cannot be sure. She has a perfectly astonishing talent for putting words together in such a way as to make successful inquiry into their intention impossible. She generally makes us uneasy when she begins to tune up on her fine-writing timbrel. It carries me back to her Plague-Spot and Poetry days, and I just dread those: “Into mortal mind's material obliquity I gazed and stood abashed. Blanched was the cheek of pride. My heart bent low before the omnipotence of Spirit, and a tint of humility soft as the heart of a moonbeam mantled the earth. Bethlehem and Bethany, Gethsemane and Calvary, spoke to my chastened sense as by the tearful lips of a babe.” Page 48. The heart of a moonbeam is a pretty enough Friendship's-Album expression—let it pass, though I do think the figure a little strained; but humility has no tint, humility has no complexion, and if it had it could not mantle the earth. A moonbeam might—I do not know—but she did not say it was the moonbeam. But let it go, I cannot decide it, she mixes me up so. A babe hasn't “tearful lips,” it's its eyes. You find none of Mrs. Eddy's kind of English in Science and Health—not a line of it. About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Mark Twain (2004). Christian Science. Urbana, Illinois: Project Gutenberg. Retrieved October 2022 https://www.gutenberg.org/cache/epub/3187/pg3187-images.html This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at , located at . www.gutenberg.org https://www.gutenberg.org/policy/license.html

She wrote this little biography her great life-work had already been achieved

Stories, News & Mentions about ACHIEVE

Achieve’s Stories on HackerNoon