Desire Paths & Markets for Recommendation
Over the past year, there’s been quite a number of projects that started to experiment with ‘Curation Markets’: protocols that use crypto-economic mechanism design to curate information.
Some of the projects that went live on testnets or mainnets are Memelordz & FOAM.
Memelordz: trade dank memes.
FOAM: Curate Points-Of-Interest.
Although only a few projects have went live, there’s been a substantial amount written about Curation Markets and its primitives. Some recent ones I enjoyed reading:
Much of the trepidation, I suspect in pushing live, is the worry that the chosen design parameters might not yield itself to the desired outcomes. In simple primitives, there’s already decent amounts of complexity.
You can see this in variation of designs of token bonding curves for usage in business models (by Achill Rudolph & Logan Saether):
Paul Kohlhaas drew considerations to consider for utilising token bonding curves for desired outcomes: what it issues, what the supply is, what collateral is used, what asset it represents, what function the curve is and how the pricing is influenced.
In token-curated registries, there’s even more knobs to consider.
The possibilities in the design space is a meaningful concern.
These “games” have hard-coded and cryptographically binding rules that are often times not forgiving or unable to be reversed (unless explicitly programmed to). These hard and fast rules are what gives them value in their newfound ability to coordinate in new ways.
We want to design these protocols such that they serve the users, and not the users serving the protocol. The latter can come to exist without intent & could even potentially be dangerous.
For example, consider that Bitcoin is incentivizing a huge of amount of energy to be burned to produce locks to secure a distributed ledger. In another example, especially in the context of information curation: if users start serving the protocol we might end up with junk & spammy information, rather than meaningful information. Or worse: viral information without substance over quality information that’s novel.
These protocols form new focal points for people to coordinate around, but if the focal point falls off a cliff, you don’t want the people around it, like lemmings, to also falter over along with it.
The power of rules and focal points aren’t new however. We often get stuck on them, because we got used to them. We become entrenched, simply because we’ve spent the energy to learn how it works. The rules become desire paths: we follow them because others paved the way.
On top of that, sometimes it feels like more complex rules are adopted in certain parts of society merely because the additional coordination required is actually a form of eustress.
In other words: more complex rules actually brings about more coordination, because in order to correctly use them requires communication. Requiring more communication produces positive externalities amongst those involved in the process. It’s like figuring out a hard problem together, as a team, bonds the team.
It’s hard to go from a familiar system with existing bonds to a new system that is simpler. Glen Weyl & Vitalik Buterin discuss in their article on the trade-offs between simplicity & familiarity.
“Focusing on familiarity (ie. conservatism), rather than simplicity in some abstract mathematical sense, also carries many of the benefits of simplicity as we described above; after all, familiarity is simplicity, if the language we are using to describe ideas includes references to our shared historical experience. Familiar mechanisms also have the benefit that we have more knowledge of how similar ideas historically worked in practice.”
So, I understand that there’s trepidation to launch these new crypto-economic systems, especially when their value comes from them not being manipulatable. It becoming entrenched around sub-optimal focal points might indeed cause the people to fall over the cliff along with it.
Thus, in the short term, I’ve been exploring the premise of utilising Curation Markets as Recommendation Markets.
Markets For Recommendation
In this variation in mechanism design, the aspect of using crypto-economics to curate information serves to produce recommendations, rather than it being regarded as the sole source of truth. A good example, is a playlist.
Applying a curation market directly to a playlist might lead to the most viral playlist, not the most nuanced playlist.
A better alternative is that the market produces recommendations for the playlist owner to choose and add to it. This means that market participants have to take into consideration the protocol and the addition of a human on the other side in order to earn from the curation.
What then becomes necessary is a participant that takes from the pool of recommendations and thus through that action verifies the usefulness of the market. It’s not assumed that they have to take the top ranked item from the curation market (or to take from the market at all). Ocean Protocol uses a somewhat similar design, called “Curated Proofs Markets”. There’s a market for predicted popularity, and a verification system of what’s actually being used from the popularity market.
This participant can be a human, a group of humans, a committee, a council, a program, and eventually it won’t need to be anyone if the curation market for recommendations is good enough that ends up serving the needs of the users precisely. In that circumstance, the curation market itself is then the truth.
An example of this in the context of token-curated registries would be something like Civil’s council.
A TCR would add and remove items from a list using its crypto-economic games. A council would select already vetted members and do a final check to determine if the market produced the outcome it wanted.
Having human (or other) stop-gaps does mean that social scalability is hampered somewhat. It doesn’t allow for grander coordination systems that extend beyond what’s currently doable by humans and it does open up the system for more abuse.
If we accept that we allow more meatspace interference in order to learn from them, utilise them and explore how they can be useful and meaningful, we can start with curation markets much sooner.
If the market for recommendations is not producing meaningful recommendations it can gracefully disappear without impacting the goal of curating novel information.
Conclusion
A desire path doesn’t demand that you walk on it. But, because others have paved the way, it’s a pretty good recommendation to follow. Unconsciously, like a stigmergic ant colony, we leave these traces in the world.
In building curation markets we don’t have to lay down the asphalt just yet. Let’s experiment with more flexible crypto-economic curation markets that allows us to learn from where others have walked.
By viewing curation markets not as the truth, but as a market for recommendations we can have these protocols serve the users, rather the users serving the protocol: starting tomorrow.
