This post is the second of a series where I aim to explain how fathom-credentials are meaningful, even though the individuals involved in creating them have no reputation, how this can lead to them not being "true" and why that is not a problem but a feature. For the first post, click here
In the previous post of the series I stated that trust in a certificate can come from two possible sources:
I concluded, that institutions have to rely on their reputation simply because processes can not be verified at scale.
Enter blockchains. Blockchains provide a common source of truth and also make it possible to create processes, called smart-contracts, that always execute exactly as designed. With no downtime. Ever.
I want to say it again, differently: We can now built technology that does not belong to anybody!
Fathom tries to create a process that can be trusted without having to trust the individuals executing it!
Briefly summarized, fathom implements a betting game in which a jury of assessors judge an applicants skill, while backing up their judgements with money. If a result is supported by a majority, the pot of money is redistributed such that those assessors closest to the final result are rewarded the most. Importantly, collusion among assessors is difficult to execute, due to how the voting mechanism is architected: In order to collude, assessors must share their votes with each other; but by doing so, they are sharing information their fellow colluders could use to betray them and instead steal their money.
Put differently, the fathom protocol is a mechanism for individuals to stake money to compare their subjective impression of someone's "mastery of Z" -- and be rewarded if their views align.
The idea that the members of the jury (assessors) are invested in the quality of the result (they have "skin in the game") is what lends credibility to the process: If I show you a certificate, it is not important who the assessors were that created my certificate. What counts is that they have risked their own resources and they did not loose them because the my proficiency was so obvious that they all came to the same result3.
This principle of staking money in a coordination game is referred to as a schelling game and it is important to acknowledge that the resilience of such methods against malicious acting has yet to be proven.
Trusting a process is much harder than trusting a person/institution, so let's have a closer look at some of its elements:
Let's get to the most important point right away:
Aggregating subjective impressions is no recipe for finding the truth! In fathom, assessors for a new assessment are those who have earner it earlier. In case of a new credential, where there are on earlier certificate holders, assessors are drawn from related credentials. And in the beginning, the first assessors will just be some trustworthy individuals.
This means that the initial assessors and the first users will have a very large influence of what it means to truly 'know Z'2. Galileo would not have gotten a fathom physics-credential!
This is very important to acknowledge, so I'll state it again: The fathom-protocol is not for finding truth in the scientific sense. Instead it allows communities that have certain standards (may they be scientifically accurate or only coherent from inside a certain worldview) to map those onto a set of credentials.
Most algorithms or processes have parameters and so does the fathom-protocol. Let's look at how they affect it (and the truth)
So the fathom-process of creating credentials is not a silver bullet that will replace higher education. It is limited in it's objectivity and the scope of knowledge it can assess (e.g. cutting-edge scientific). This is especially true for the beginning.
Yet, there are ways to address this and those are what make the protocol so interesting:
If for example, 3 out of 5 anonymous assessors (you would only see their ethereum addresses) would attest that I have a solid knowledge of the intricacies of XXXXX, you would have all the reasons to be doubtful.
Yet, if it were not 3 of 5, but 7 out of 9 assessors and some of those seven would have web 2.0 profiles (e.g. LinkedIn, Twitter, GitHub, etc) verifiably3 attached to their addresses, it would be a different matter.
Additionally, there is the price that can be adjusted. If an assessee put's up a lot of money, he can expect the assessors to go through greater lengths for verifying his skill and maybe ask them to provide some additional material besides just logging in a good score.
Also, as an outsider, I would be more likely to trust a verdict if the assessors involved in it were willing to put up large amounts of money, thus signaling their confidence in coming to the same conclusion as a bunch of other assessors they didn't know anything about.
Finally, let's look at what this variability means for the reach of this project:
As anybody who can define a piece of knowledge solidly enough to allow individuals to assess it via internet can create a credential and anybody willing to spent time can participate as assessor, there is no limit as too how many persons can be assessed and how many subjects can be represented in fathom-credentials at the same time.
This is a stark contrast to the current certificate landscape, where certificates have to have a reputation before they can be meaningful.
In their minimal form, fathom credentials place a strong focus on subjectivity, so that believing a fathom credential does not mean "I trust that this is the truth", but rather "I see that X people believe this to be the truth and they were willing to back up that belief with money".
Being able to interpret fathom-certificates will not be as straightforward as conventional certificates, but it will be the same for all kinds of them. Whether it is a hands-on technological skill or knowledge of a theoretical concept.
In the last and final post of that series, I will take a step back and place fathom-credential and insititutional credentials next to each other to see where they intersect and where they are unique.
1 For a more elaborate explanation of the redistribution-mechanism see our whitepaper or this explainer post.
2 We spent some time thinking about how to facilitate disagreement and allow newcomers to create competing definitions and put methods in place to encourage better defintions (better here means more helpful to a community)
Of course, my certificate becomes more credible if you can learn who my assessors were, where they work, etc... The point I am making here is that as long as you can be sure that there are different from me, it is not strictly necessary condition to being able to trust the certificate.↩
Although we hope that this will only be necessary when initially establishing credentials & communities and that it won't be necessary once the network has scaled beyond a certain size.↩