Lets begin with the summary conclusion.
A blockchain token* serving no technical purpose can enhance social welfare by serving as a low cost, unrestricted, always open, international and regulatorily compliant casino for funding investment in public goods.
The goal of this piece is to rebut the following seemingly reasonable claim
A project which does not require a token to operate should never introduce one.
This rebuttal will attempt to offer a plausible counterexample, in the very specific sense that the option to introduce a token enhances social welfare. There are of course projects that do require a token of some kind to operate, but I will leave that discussion for another time.
Lets us begin.
The following premise is crucial.
The price of some tokens does not rest on a rational appraisal of demand for current and future use, but only current and future demand from other investors.
Now consider a project, software based or otherwise, which does not require a token to operate. If such a project introduces a token, in the banal sense of just launching it with some peripheral integration with the project, then any sustained market price must be of the form above. Moreover,
a token preallocation for the project would increase the quantity of funding for the project
In principle, voluntary contributions could be depressed due to crowding out, but I will take it as a given that that is not the case, as seem confirmed by all practical experience so far.
In these markets, returns are not compensation for supplying productive capital, but rather zero sum losses of other participants, but keep in mind that
all losses among participants are offset by gains to others — including the project manager, as a result, these are pure transfers with no impact on the efficiency of the introduction of the token.
To raise social welfare, we need the project preallocation to deploy resources in a way which is an improvement from before.
If the project is useful and poorly monetised, then social value of the marginal resources invested must exceed costs.
Now, the only remaining question is, how can we be sure that too much resources won’t flow into the project? In the above model we cannot, but considering projects with sufficiently large capital requirements easily offer us cases where this is not a plausible problem. Moreover,
since poorly monetised projects often generate non-rivalrous goods, the generated benefits are often so large as to warrant very large optimal investment scale.
FoldingCoin is a Counterparty assets paid out to contributors of computing power to medical and scientific projects that are part of Stanford’s Folding@home.
Importantly, this project produces basic scientific data, where the beneficiaries of this data are not required to hold or pay tokens to enjoy the benefits of the data. As a result,
all demand for FoldingCoin is purely speculative.
There are currently 381,000,000 FLDC in circulation, all paid out for such computing resources. At the current price of $0.019879, that means that
$7.5M worth of computing power was attracted to this project, and in an expected sense, this was almost certainly welfare enhancing.
As noted above, returns in these token markets are stochastic zero sum. While participants may have a different expected return, depending on a number of factors, what seems very clear is that just about everyone in such markets are aware of the musical chairs aspect of the activity.
For all intents and purposes one can think of participation as a form of gambling, where people are motivated by a mix of aspirational and recreational objectives.
From this perspective, it becomes easy to see the productive social function of this token market is an extension of some existing and well established practices.
Both in the public and private sector, billions of dollars are raised annually for public good(ish) causes through lotteries, raffles and gambling.
The fact that one clearly does not need a blockchain to run a token systems begs the question of their significance in this context? First observe that
most jurisdictions have severe restrictions, or even government monopolies, on a wide range of gambling activities.
Even when they are permitted, there will be heavy restrictions on scale, form, participants (age?), opening hours, oversight and license fees. Worse yet, these will vary from country to country.
The cryptocurrency ecosystem is simply the lowest cost means of operating an international public benefit casino.
The blockchains themselves are only part of this efficiency, through their unrestricted international value transfer, but trading platforms and on & off ramps are also critical components.
You would be forgiven for thinking this is all a bit strange, but its long been known that when you start outside of a first best situation, for example when dealing with public goods, then doing all sorts of dubious things becomes virtuous.
In economics, the theory of the second best concerns the situation when one or more optimality conditions cannot be satisfied. (…) if one optimality condition in an economic model cannot be satisfied, it is possible that the next-best solution involves changing other variables away from the values that would otherwise be optimal. 
The above model seems to withstand most major critiques so far levied against tokens.
First off, note that this is entirely immaterial to the question of efficiency.
Second, defining what is or is not a Ponzi is notoriously difficult, and so is also distinguishing it clearly from gambling. With respect to the normative dimension of the activity. What does seem material to that concern is whether participants are informed about the nature of the activity they are engaging in, which has already been addressed.
A deeper inspection of most poorly monetized protocols in existence today will likely yield insights about how there were particular circumstances encouraging investment in R&D, and that there is a large survivorship bias one has to take account of. More fundamentally, one cannot escape that when something requires costly resources to develop, and has poor monetization, this will manifest in one or more wasteful ways, causing a project to
This depends on what is actually driving current activity. If people are, incorrectly, buying tokens because of poor forecasts about their future viability, then yes, that cannot last. However, if a large component is explained as gambling, then that is an activity which in principle is sustainable.
By goodness, we here mean the extent to which the project generates public goods. This is in principle possible. Since the prices that one project pays for factors, such as blockchain developer talent, do not cover opportunity cost of using those resources in other public goods, we cannot be sure there is no crowding out. There are a few reasons why we should not be so worried about this.
Its easy to appreciate the difficulty of identifying any redeeming characteristics about the contemporary token ecosystem. Most token projects are badly conceived, raise too much and have poor execution. But perhaps even worse, the arguments invoked to justify the social value of tokens are frequently vague.
The above model is perhaps not as flattering as proponents would like, but at least it gets the job done while being precise and coherent.
*By token, I really mean any digital asset on a decentralized ledger system, so it covers Bitcoin, Ether, Counterparty asset, ERC20 token, etc.
This is a hypothesis in search of review and criticism, and not my decided position.