On the Utility of Hackathons — A Fantastic Story

Dawid G.
5 min readJun 13, 2021


Photo by Dan Asaki on Unsplash

Imagine this: you are part of a tribunal. Your task is to decide who, among a population of humans who dare to face your tribunal, gets to live in peace and harmony and who doesn’t and is sent back to toil in the squalor of the unworthy masses.

These humans are selected by your tribunal according to their performance of a song to appease the scary, all-powerful ruler that governs the entire land. Yet no one, including the tribunal, has ever really seen or made sense of the the nature of your ruler, outside regular visits to the dark cave from which his voice emanates. All live in fear and awe of him, including you and the tribunal.

It is a high-stakes game for the poor sods who petition your tribunal. If they sing well, as per the judgment of the tribunal, they will be allowed to join the sacramented few who are safe from the unpredictable wby the ruler. If their song is not to your liking, they will be condemned to further toil in the squalor of the common humans. Those who do well get to join, those who do badly get nothing except having incurred a loss w.r.t. their fellow humans who chose to toil away to ensure their family’s continued existence.

The drawing is upon the land: the tribunal must soon decide who gets to join the sacramented few and who doesn’t. The tribunal’s constituents do not know how the song should sound like — they have memory of their own song and how that allowed them to join the tribunal after many decades of service. And they have experience listening to others’ songs — but the ruler never gave clear indication of what actually pleases him. That remains a mystery, but the the process must continue lest all fall under the ill will of the ruler.

Should the losers lose?

At first sight it appears that the highest utility would be for each commoner to try as hard as possible and as many times as possible to join the sacramented few— facing the tribunal is free of cost and decisions are swift. Yet, no-one knows how to sing — it is a wild guess as there is no feedback except the binary yes/no.

There are different streams among the commoners: some profess singing loud and croaky, others sing mildly and intelligibly, another sect hums softly without words — among dozens of others. How to act? The cost involved with following any stream is significant: to spend a fraction of a lifetime in preparation is the price. What strategy bears a higher chance of success — random attempts based on rumours or a strategic decision based on education?

Can this be a fair game? Those who happened to apply themselves to any of the existing streams incur a high opportunity cost: they learn a skill that has no value except when they are selected by the tribunal, in which case everything changes. Given that nobody, even the tribunal, knows whether the contributed songs are of any overarching importance, nobody is actually competent to judge.

Nobody is actually competent to judge.

In technical terms: the decision function (the tribunal) is based on the statistic the members compile off the signal that the cave provides (in case you missed the hint: gale winds provide the randomness for the voice-like sounds the cave emanates — the cave is empty and the ruler a spectre). The singers need to approximate the decision function as best they can, while the tolerated error between the two is unobservable to anybody except the tribunal’s members. Can this mechanism, arbitrary as it seems, be fair?

It is fair at its beginning — when no tribunal exists. After the first tribunal comes into force, the game begins to become more and more skewed, putting incumbents, who cannot be objectively competent, at an advantage as well as those among the commoners who are hoarding or peddling their “knowledge” of how to approximate the decision function. Both are still forming something out of pure randomness — “ex nusquam”, something out of nothing.

The incumbents and the peddlers have mechanisms to reimburse them for their costs. The losers get only costs. If your gut reaction is “by my grandfather’s beard — that is as fair as it comes!” ask yourself this: what do those who attempt all their life and never attain the sacrament of the tribunal get in return for their efforts, other than bitterness? Remember that they did not learn anything about the decision function or how the error is measured.

Hackathons don’t work this way, sir…

If you attended in-person hackathons when that format still dominated you might have noticed that somehow, in some way, everybody won something. A great experience, everybody winning something, from small to big. Free food. Learning what does and doesn’t float their boat in software engineering, business ideation or general human collaboration.

Everyone comes out ahead. And that is good.

Remote hackathons, especially the selective kind, run a risk of breaking this unspoken social contract: if the budget assigned by the organizers is spread only among those who contributed in a way the tribunal recognizes, aka. “Y k $ in prizes for the best solutions”, the ones who try and prepare and fail are left with little to show for it except bitterness. And bitterness and a swath of other human emotions are destructive if there is no help in sensemaking of the outcome.

In the age of magic internet money it is easier than ever to ensure that no participant is left wanting, falling prey to their own destructive thoughts and emotions. By including even the smallest token of appreciation (no pun intended) making sense is made easier for those who did not make the cut — while still providing a strong learning signal for everyone involved. The net outcome is a win for everyone.

The more expensive route is to engage in dialogue before, during and after a remote hackathon to provide all those whose contributions weren’t rewarded with help to make sense of their individual outcome and forge a path forward. This finesse however is pricey in both labor (community management) and fixed costs (keeping the infrastructure online longer).

Hackathons are just one example. The management of contributions to DAOs is another, as is the path of joining decentralized worker collectives.

Embracing non-extractive purposeful communities as self-regulating mother processes that deliver outcomes for those who spawned them as well as those who align with them seem to be the future. That does not mean that the “tribunal game” is defeated — but it creates a safer space to address this issue.

Arguable examples of nearly-non-extractive purposeful communities:
About Gitcoin — Gitcoin’s Blog
Quick Start Guide — Deep Work Studio
… do you know more?

If you believe feedback improves lives (human and AI alike)— you can provide some here:

Feed the machine with feedback!



Dawid G.

Helping build sustainable economies. ML Engineer and Token Engineer, MSc CS and Intelligent Systems Engineering + BSc Biomedical Engineering @TUHH @Todai