Twitter sucks for meaningful, threaded discussions.
That said, today I had a really interesting discussion with Vinay Gupta, Ian Grigg, Clockwork, Wulf Kaal, Marco Aniballi about decentralized governance that started here with reading this paper:
- Blockchain Infrastructure for Measuring Domain Specific Reputation in Autonomous Decentralized and Anonymous Systems
Authors:- Craig Calcaterra
Department of Mathematics, Metropolitan State University - Wulf A. Kaal
University of St. Thomas, Minnesota - School of Law - Vlad Andrei
HighTechBlock
- Craig Calcaterra
It's a 44 page read, but well worth it for those interested in decentralized governance systems. About half of it is examples.
There's also this essay by Wulf:
Clockwork chimed in mentioning work by Ian Grigg and an interview with Vinay Gupta:
Ian replied with:
Your reputation system can never be better than your identity system. Solving the sock puppet problem with tokens is not solving an identity problem, it’s applying first order discrimination (marketing term) to a social problem, and gives equally unsatisfying results.
I’d love your thoughts on Wulf’s paper. I know 44 pages is a lot to ask, but half of it is examples. I’ve already emailed it to Vinay. He and I have had some conversations about these issues already. I think Wulf’s approach may avoid the dangers he’s rightfully concerned about.
With the idea of contextual tags with staked reputation and broad payouts across all reputation holders for the tag, he argues you can solve sybil and sock puppet without actually needing identity. Quite a bold claim. Co-authored with a mathematician. Curious what you think.
Ian: can you clarify first order discrimination? When I search I’m getting first-degree price discrimination results.
Ian clarified:
yes that’s it - first degree price discrimination is, in short, to discriminate for marketing goods by charging a high price for one and a low price for another. It works, but it is so brutal and clumsy a discrimination that it leaves money on the table.
From the way Twitter threads things, it's hard to realize there's a much deeper conversation buried here which includes:
Ian:
WoT in the PGP form failed for two reasons - 1. lack of standards, so the edges in the web were easy to misinterpret; 2. no handling of liability for adverse results.(CAs/PKI fixes the 1st error but not the second.)
https://wulfkaal.com/2020/05/25/self-sovereign-identity/ is looking for a technocratic solution, but attackers can use the same technical approach.
Instead, need to think of an asymmetrically harder approach - one that is easy for ‘good’ people but harder for ‘bad’ people. Eg social based, skin in the game.
It gets mostly lost, but you have to drill down here to see the full discussion.
Marco:
Properly implemented, sovereign identity eliminates sock puppets (sibyl). Anonymity has always been a luxury derived through systemic inefficiency. It is not sustainable IMHO since it removes accountability. Privacy, on the other hand, is a manageable function of an ecosystem.
Me:
I agree that being anonymous and in some ways having privacy at all is a relatively recent idea for humans. I wrote about that a few years ago: @lukestokes/privacy-identity-and-human-flourishing
Paradoxically, I think we need both privacy and radical transparency for a functioning free society.
Vinay:
In almost all cases, we want personal information about people so we can estimate or reduce their risk of not paying their debts
If you provide alternate mechanisms for paying those debts - insurance, escrow and similar - you don't need identity http://guptaoption.com/4.SIAB-ISA.php 2006 work
We live in a hostile world, and we are all prey animals relative to the State and Society. Sometimes we get farmed, sometimes we get hunted, we all have secrets, we all have something to hide, because as society changes, your perfectly safe, normative position can be demonised.
Then you have to dig here for more back and forth with Vinay and Marco:
Vinay:
We live in a post-Holocaust world.
Computers were essential to conducting the holocaust https://en.wikipedia.org/wiki/IBM_and_the_Holocaust
The world is filled with persecuted groups: black people, gay people, gun owners, Christians
All identity data is bad for the individual and their rights friends
Marco:
PUBLIC identity data is bad. Private verifiable identity data is good.
Vinay:
Always, always, ask yourselves "if this system had been running great in 1920s German, how would the Nazis have used it in the 1930s?"
Wow, your reputations systems - all of them - don't perform so well when people get downvoted whatever they do just because they are Jewish.
Marco:
One of the several questions we asked ourselves in designing our models. Similarly, though, you can't stop people from "picking on others." Identity infrastructures must allow society to choose. In decentralized ID scenarios, 1930s Germany would have started and stopped faster.
Vinay:
Focus on what matters: "if I transact with this person, and it goes wrong, will I be made whole?"
If the answer is guaranteed yes, I will be made whole what do I care about the identity of the person?
This still leaves "criminal" risks - active malfeasance.
Subpoena the ID.
Marco:
You care about the identity only insofar that you know the person you are dealing with is uniquely accountable. Otherwise arbitration becomes necessary and the system fails. PII transfer is not required for accountable identitification.
There's a branch off reply thread which starts here
Vinay:
And in that instance, I can take the insurance policies from two different insurers, and ask them to perform a ZK check that these policies do not refer to the same person.
Then, of course, the question of "can I trust my insurance intermediary."
Well, that's where the fun is.
Marco:
These are all problems that can be solved by identity. The nature of identity is the real issue, not whether or not it is needed - it is. It undermines most forms of white collar crime and simplifies interaction between stakeholders. Tx insurance is just another band-aid.
Vinay:
I don't think that's true. I think identity is just a route to compensation when things go wrong, in almost all commercial transactions.
Criminal activity, where the objective is to punish, does require identity obviously. You need to be able to get the body.
Clockwork:
So essentially, real world systems require identity as a preemptive measure to minimize the costs of obtaining it AFTER an activity (such as criminal) requiring it has taken place.
Marco:
In a world where everyone (individual and entity) is uniquely, privately and immutable identifiable, commercial recourse is a remote edge case. Imagine the surveillance "state" where everyone can surveille everyone without PII and PII is never disclosed.
Vinay:
Ah yes, the "bottom up surveillance state" which I have always dreamed of living in, where everybody watches everybody all the time - it's like living in East Germany - but we can't overthrow the State because there are no secret policy headquarters to burn down
This is not good
Marco:
Think deeper on that. Lying in that type of world is very difficult. If there's nothing to burn down, what happens?
Vinay:
People remain trapped in a surveillance society forever.
Because it's not just the thieves that get downvoted, it's the minorities, it's the gays, and the blacks, and the communists, and the women, it just depends which prejudices your society has.
What you are creating is hell
Marco:
Accountable hell. (see Sartre). By the way, no one gets to vote indescriminantly. Voting happens by activity, choice, decision and other transactions. You don't get to down vote a gay counterparty unless you interact with them and they perform badly.
Vinay:
I think by the time you're recommending a scheme for creating a surveillance state - of whatever kind - you've gone far, far from the path to an acceptable future.
Privacy and freedom, not decentralized surveillance, please. You are not proposing a future we should live in.
Marco:
Surveillance state was a metaphor. We live in one now! However, it is one-sided. My proposition is that of an accountable society where "intelligence" is available to anyone, not just power structures.
Your discriminated attributes are yours to be disclosed (or not) at your leisure/discretion. Parties requesting those disclosures do so beneath the all-seeing eyes of everyone. How long would would an entity get away with requiring disclosure of sexual preferences?
Or religious affiliation?
Another branch off here:
Me:
The idea that people value this information and secure it rationally based on the probabilistic threat it will be used against them doesn’t hold up.
People give this stuff away for free in droves on social media for memes and cat pictures.
Marco:
Share away on FB knowing that it won't be tied to your work profile, unless you disclose the link there explicitly. And that said, full accountability is something we've had little familiarity with as a society. There will be a learning curve.
Ian:
it’s just a data science problem to correlate a personal profile to a work profile.
Given that, is it even responsible to propose such a thing?
Marco:
It does if your identity descriptor is different for every interaction.
Ian:
Welcome to China.
Vinay:
This is 100% the wrong approach on every level. You are not describing a better state of affairs, but a worse one, where rather than being watched by 2 people, we are watched by 2000, and call it "democracy."
No no no no no no. NO!
Marco:
Scary, isn't it? But it isn't democracy, it's accountable anarchy. Something few in today's world have the stomach for, but long term a solution to so many of today's systemic failures
Ian:
It allows for perfect bullying. We’ve already seen that on the Internet. To propose it for all society is a horror.
You won’t be a person anymore, you’ll be your public reputation. Your identity will gone, shared, spoiled, eaten.
Who wants that?
Clockwork:
Not that I agree with Marco, I don't, but playing devil's advocate, bullying on reputation only without PII is not as easy as it seems. Trolls and online bullys usually grasp onto irrelevant PII to do their "job"
Marco:
Trolling becomes risky when there's no anonymity. 🤣🤣
Ian:
not really. this is the point about anonymity versus lack of liability - revealing the person doesn’t mean you can punish them.
Cf British officials are currently flouting lockdown rules - we know precisely who they are, but what happens?
The question isn’t ever - as Vinay was pointing out - who are you?
The question is, what happens if you do something wrong.
Clockwork:
also, what happens to those who enabled you? Should anything happen to them at all? And in a reputation-based system, what happens to the people that gave you your rep rating etc. etc.
Vinay:
Literally the only exceptions are poker, auctions, and insider trading. (ok, and maybe a few other things like that)
But really identity hardly ever actually matters.
Clockwork:
that too. However, your original premise still assumes the same intelligence/information/"power" given to everyone which fails like every other "democratic" structure. As @iang once said, "If we let anyone play then … we get everyone"
Marco:
It's isn't given, just available. It isn't democratic, it's anarchy, with accountability. However, democratic systems could use it too. It doesn't solve our socio-political issues, just provides a better platform to execute them with.
Now back to the main thread:
Vinay:
I think it can and should be taken further: if I have a one-use ID for an insurance policy, and proof that I can claim on that policy if you misbehave, do I need identity at all?
Really only in special cases like poker games and auctions, where I need sybil protection.
Marco:
Ask FATF if you need identity for insurance. 😜😜
Ian:
Ask FATF why you need identity for anything and you’ll get crickets. They can’t solve anything, they can only double down on what already is shown not to work.
So why did I bother to copy all these tweets here on Hive?
- I think conversations on decentralized governance are really, really important. Violent centralized governments are out of control and we need competing systems for society to choose. See democide.
- Twitter is a mess. Other than the people on this conversation, it's very difficult for others to follow what's going on.
- The people involved in this thread have, in some cases, 20+ years of experience each on these topics. They are some of the best in the field, and we get to hear their thoughts in real time! How cool is that? This is the true value of effective social media.
- Hive is immutable. Because this conversation is so difficult to follow on Twitter, I was concerned at one point that tweets had been deleted because I couldn't find the thread. That's really frustrating when there's time and energy put into a conversation that can be really interesting to others, but all of that is gone if tweets are deleted. I've had similar things happen on Facebook, and it's frustrating.
If you're into governance conversations, I hope you found this interesting.
Reputation, identity, and decentralized governance systems are really important, but if we don't build them correctly, we may be building tools for our own enslavement even worse than what we have today.
