Whuffie - how to build a reputation currency in Ethereum?

guakaguaka Member Posts: 18
Down and Out in the Magic Kingdom[0] has proven to be a defining book for my life. It introduced me to the concept of Whuffie[1], a reputation currency around the time I was volunteering for CouchSurfing and hitchhiking all over the world. I'd love to work out a system that works in a whuffie way, to indicate social capital. So a currency that doesn't decrease when you pass some to someone else.

I guess the main problem here is a sybil attack. How to avoid people creating multiple profiles?
I already mentioned it here: http://forum.ethereum.org/discussion/461/basic-income

For now I just want to put the idea forward. Let me know if you're interested in building this.

[0] http://www.craphound.com/down/download.php
[1] https://en.wikipedia.org/wiki/Whuffie

Comments

  • guakaguaka Member Posts: 18
    edited February 2014
    Hi Karl, 7 month ago I've asked Cory what he thinks of Bitcoin. He was skeptical. Yesterday asked him over twitter if he changed his mind in the meanwhile, no response (yet?). Of course he might think differently about Ethereum though.

    If it's possible to create multiple profiles that can give out reputation it's very easy to game the system. Take a look at quora, it would be very easy to get millions of points just through some bots and humans solving captchas. StackOverflow would already be a lot harder to game but it has a much higher point of entry, which would defy the purpose of a BitWhuffie system.
  • Karl_SchroederKarl_Schroeder Member Posts: 37 ✭✭
    I see your point. I think we're approaching the concept of reputation from different directions (which is exactly what happened with me and Cory when I was reading Down and Out in manuscript), but that's okay. I say run with it. Cory's got a very particular perspective after years of hammering away at the DRM and copyright issues (more power to him), and the block chain is a very complex beast; I don't know of any SF writer who's caught up to it yet (a terrible statement about how current we really are). I'm still trying to sort out the differences between the BC as a rights-management enabler as opposed to an enabler of verifiable assigned digital identity. These are actually different concepts in significant ways: the former is about the content, the latter is the creation of a digital "thing" whose actual content is not as important as the fact that it behaves *like* a physical thing (has persistence, singularity and particularity etc.). The BC enables us to assign certain intuitive notions about "thingness" (or more exactly, assign certain metaphysical attributes of object identity) to digital entities. This is fundamentally different, in my mind, from "controlling access" to a piece of data, which seems to be the aim of DRM.

    Of course, I may not have caught up to what's really going on here either.
  • StephanTualStephanTual London, EnglandMember, Moderator Posts: 1,282 mod
    Potential solution to Sybil attacks:
  • StephanTualStephanTual London, EnglandMember, Moderator Posts: 1,282 mod
    In a nutshell, make identity expensive by creating deposits that both the user and service provider cannot cash in - then are automatically returned to sender after say, 6 months.

    After 6 months the user can decide to extend the deposit. If the user is a bad actor the service provider can terminate the account on the off-chain side or negotiate an increase in deposit.

  • SofSof AustraliaMember Posts: 10
    Clever approach.
  • SatCaSatCa Member Posts: 29
    I understand a lot of us are trying to do some really good work in attempting to create an honestly Good International Economy...

    I understand we're working on a very powerful concept, I believe this should be done only once and it should be done right. If someone can do it better, I believe we should move out of the way and help.

    The issue of prime importance is enough 'maturity' (In terms of EQ, IQ & Spiritual Quotient) within initial members.

    To solve the issue, I suggest we play a game. Vitalik referred Go (the game) for proof-of-distribution.

    I have a different game in mind. I have already begun playing the game, no one has secured the first spot yet, so I'm still waiting for that.

    I would really love to share more about the game, but I need a central station.

    For now, I'm awaiting responses, queries & players:

    http://honestcoin.createaforum.com/general-discussion/

    Hope you could find some time to engage me in a discussion :)

    To solve the human problem, you need those who understand humans.
    To solve the economic issues of the day, you need trust.
    To solve the problem of understanding human, you need love
  • lightcoinlightcoin Member Posts: 6
    This sounds like an interesting project that I would like to see implemented as well. I wrote a blog post a while ago

    http://p2pconnects.us/2013/11/15/finding-my-unicorn-a-decentralized-reputation-rating-system/

    and subsequent forum post on the Ripple forums

    https://ripple.com/forum/viewtopic.php?f=1&t=5081

    brainstorming how it could be done. If anyone can implement this, I'll help beta test :)
  • JasperJasper Eindhoven, the NetherlandsMember Posts: 514 ✭✭✭
    The 'locked deposit' that is >= in size of the amount of coin in a 'bank' isnt neccesarily something that 'identifies the human', it is just a mechanism against abuse being profitable.

    Still, it is a good first step, and a good tool in the bag, and already in a todo list.* but can only become an identity if something goes on in the system where people identifying identities as being sockpuppets has consequences. (psuedonyms might of course be tolerated)

    @SatCa what game? To be honest, we already have fora..

    One 'gamecoin' is huntercoin. The problem is that the incentive to create AI-players goes up terribly when a coin is serious, and effectiveness of the game to test for 'is human' goes down terribly. Besides, it focusses on waste of human time like PoW wastes power.

    You could try use a web of trust, but have more kinds of activity contribute to it. People can help reinforce 'evidence-for-human' in more situations. For instance, if others have trust you are some particular psuedonym/person, and you buy a good/service and indicate it went well, that indication could potentially be used in a web of trust. A requirement is that the store goes under a particular name that is associated with a particular public key. i.e. stores cant sell out 'web-of-trust trustability' to third parties. I think this can essentially be done by having software pay to names, not identities. Changing the public key of a name a lot might be punished.

    (going off topic)The video starts essentially about the transaction volume problem. Still kindah expect it is possible to tie a lot of blockchains together, and then people can be full-node in some combination, and lightweight in the rest, so large datacenters are not needed. Merged mining? PoS-is-cheap? http://www.reddit.com/r/ethereum/comments/1z1ftt/mt_gox_failure_and_opportunity_comments/cfqx7mn That said, maybe i am being naive, i should look at it more closely..

    * https://forum.ethereum.org/discussion/639/servers-being-paid-by-scripts-owned-by-viewers
  • SatCaSatCa Member Posts: 29
    @Jasper Bro, sorry been busy.

    I agree "evidence for human" is a requisite. My take on this project is totally different.

    The understanding that faith gives money value, is an old theocratic secret. This was supposedly first understood in depth in India.

    People in power have used and abused the secret again and again. The Illuminati was the last group I know of to do so prior to Bitcoin.

    This is a secret that men of means share with each other and apply to get their secret groups richer, understand that the same is always at the cost of others.

    Money leads to governance, which leads to the 3rd front. i.e. Journalism.

    These men of means, currently hold the reigns to most of the bigwig news providers worldwide. They are adept in strategy and are used to playing with risk.

    While the common man decides to understand what is happening, these men I believe would have easily prepared themselves for "History to repeat itself".

    Money, governance & media all are extremely powerful weapons in the hands of skilled strategists.

    While we discuss, there are others who have been asked to position themselves well to "take back the internet". I've been planting counter - intelligence info for those who are already in Bitcoin. I fear the bankers and government are still not honest about their involvement in the same

    I believe my statements have been read (by Big Brother/s), and the political race to politely usurp control over the internet has begun.

    Those who were most prepared were able to react a lot faster. Father of the Internet is out to solve the problem, Edward Snowden is doing video interviews.

    I humbly feel, if they understand how to solve the problem. They shall fuck the internet.

    Unless one realizes, that these problems are just reflections of our problems on the ground. Creating similar systems online shall just invite these problems online.

    There are also a covert ops being carried out in the crypto world to track hostile enemies. I shall not divulge details of this project, lest I hurt the sentiments or spoil plans of some Intelligence Agency trying to carry out its responsibilities.

    I'm creating a ruckus, cause innocent people are being involved almost intentionally. If these agencies have plans of financial profiteering of off others losses, then the world must learn about this.
  • rdnkjdirdnkjdi Member Posts: 135 ✭✭
    If this had a smartphone application front end - I wonder if you could integrate it with only humanly doable tasks.

    For instance - application asks

    Would you rather run a 5K, turn your GPS on during this time.
    Meet another Whuffie > 10 reputation face to face
    Take a picture of a random stranger in front of an identifying place where you are registering from

    At this point you know the person is human (even if they are creating another account).

    Could the value of a Whuffie be partially determined by the value of the party giving it? In social situations this is how it works.

    If the popular guy introduces you - your social value is already well beyond what it is if the introvert with no friends is the guy you are interacting with.
  • happyseaurchinhappyseaurchin Member Posts: 6
    I'd put money in, showing my trust of the people and the system we are using. I would not want to have multiple registrations because I'd like to build on my reputation. I'd not use negative, since that gets problematic, and I'd use a unitless system, so that our values are 'discussable' outside the actual numerical trust metric.

    Yes, I think one of the first things to build on ethereum is a reputation system, and I'd suggest, closely followed by a moneyflow system.
  • compleatangcompleatang Member Posts: 1
    What if the proof of humanness|reputation mechanism was an endowment parents gave to their children (certified by some other humans with a low probability of collusion) which was kept for life and distributed along with other assets upon the expiry of the human?
  • aatkinaatkin Member Posts: 75 ✭✭
    I'm working on a system based on TrustDavis. In some respects my revenue sharing model is based on distributing funds based on a 'Whuffie' type app coin. We're trying to get this working on Ethereum, Codius and Firebase. Feel free to contact me for details. Thanks.
  • CerereanCererean Member Posts: 10
    I posted about this a while a go, when discussing verifying unique identity. My proposal was to tie account creation into iris scans - you only have two of those, so your opportunities to create sockpuppets is limited. Of course, that doesn't work for people with damaged eyes. Maybe something else unique to people, such as fingerprints or - casting our gaze a decade or so into the future - DNA could work? Or some other means no-one has thought of yet...

    The suggestion (from someone else) was that, every time you met someone else using the system, they would verify that you're a person using their own rig. My addition was to have a reputation system added to that, so that when it becomes, say, 90% likely you're a real person (who's not faking the signals from a thousand hacked phones), you'll be able to receive and give repcoins, and the more verifications you get, the more you can do that. Up to 100 repcoins a month allocated to each user, with a half life of one month, so your reputation on the system decays in half every month. No negative rep, because that's just a lightning rod for bullies and trolls who wish to use their monthly allowance to put someone at the bottom (though they'd be neg-repped in turn... but that just turns into a neg-rep war).

    You might actually be able to do it without anything like a fingerprint to tie it to a unique person.

    Going further, I'd like to see a more useful refinement, that would allow you to rate people both on their trust (terminal) and their ability to judge character (instrumental), in a web of trust that will allow you to get a trustworthiness score of someone who is a couple of links away on the basis of how many people trust them who you trust to judge character...
  • aatkinaatkin Member Posts: 75 ✭✭
    Our initial approach is to tie identity to a verified email address. This gives pseudonymity if you use a different ID for each transaction. You cannot build reputation that way, so you would either use your real-life email and attach other social network info to it, or you would use a pseudonym for a type of transaction (say [email protected] vs. [email protected]).

    TrustDavis gets around the Sybil/Sockpuppet problem by making references expensive. Creating sockpuppets will lose you funds in the long run. Creating poor quality reviews/references will also lose you money immediately.

    One thing this system would sacrifice is privacy, as everything is on the ethereum blockchain. Ways around this might be the OpenBazaar type GUID for a given node, or possibly some kind of HD ethereum wallet address which could only be verified by the node and a centralized entity (i.e. the Dapp).
  • CerereanCererean Member Posts: 10
    In a web of trust network, wouldn't sockpuppets show up as an anomalous "knot" in the network, where there's a clump of accounts which all verify each other but aren't verified by any, or very few, outside the knot?

    Additionally, one could tie that in by making it a smartphone app. Designing it for smartphones would also allow the poorest in the world to get into it, people who will get smartphones as their first modern piece of technology, who are most in need of a means to verify and rate people they do business with. The vast majority of people won't have more than one smartphone, and the above mentioned knots might serve as a warning for potential scammers. Sure, you'll have some people who will be determined to create multiple identities and will succeed, but would that be any more likely than it is now with forged documents?

    You could even have multiple aspects of trust involved, such as terminal and bridging trust - the former is "I trust you this much", and the latter is "I trust you to be a good enough judge of character that I can trust those you trust". Combining all that together will give you a rating of an individuals trustworthiness.
  • H3g3m0nH3g3m0n Member Posts: 4
    edited May 2015
    Reputation Economies have some fairly big issues.

    Some people are more public would make more money. You get paid for being a celebrity or making a lot of noise. Also people might get paid more for being more likeable so if your ugly you likely to get less. There are also things like gender/race issues (Are females paid more? less?). People will be encourage to just spout the hive mind opinion rather than give descenting opinions, basically it encourages everyone to participate in one big circle jerk. People will less noticeable jobs are going to be panellized. Consider they waiter vs the guy cleaning dishes in the back.

    Also people have to actually use the system.

    Firstly there has to be an incentive to 'pay' someone (ie why don't I just hoard all my rep). If there is some reason to give rep, then why not just make a pool where people give each other rep for no reason.

    Another issue is people have to actually be bothered to use it. In Cory's book they had augmented reality. Maybe icons on forums. But even those probably require a bit of effort.

    Can you give someone negative rep?

    Having said that, small ones could be useful for forums, small communities and the like. And it does make sense that some would do federation to similar communities and that meta-reps could spring up where they take the average of the id on all communities.

    Another good fiction source for it would be the creative commons Pen & Paper RPG Eclipse Phase: https://robboyle.wordpress.com/eclipse-phase-pdfs/
  • RadivisRadivis Member Posts: 6
    What happened to this project? Is it still actively pursued?

    I have developed a reputation economy system called Quantified Prestige at the end of 2011, but haven't spread the idea very far. Back then I had to struggle with serious health issues which forced me to put the idea on hold. Now I feel good again and am willing to pursue Quantified Prestige further. Still, the latest documentation of my system is from 2012: Quantified Prestige Documentation V0.02.

    Note that QP is a very elaborate system with a peer-to-peer esteem (positive reputation) allocation system at its core. Around that, a flexible trust system is used to minimize the attack surface for Sybil attacks. Finally, if a user has esteem and trust, she gets Prestige, which is a reputation score.

    Inflation of reputation is prevented by limiting the number of esteem points a user can allocate to others to a permanently fixed quota. Esteem points can be retracted at any time for any reason.

    I am aware that reputation economies can have serious issues. So, I tried to address some of these issues in the appendix of the documentation above. My stance is that a reputation economy would be an additional opportunity for all people to expand their personal wealth. The good this system does will most probably outweigh its problems by far.

    Perhaps the most promising application of QP is the possibility to abolish digital scarcity: People get paid to give away digital goods for free!

    How would that work? This requires a applications of the QP system that I call reputation currencies. Reputation currencies provide people with Prestige with a regular income in some kind of electronic currency. After some thinking, I found out that it would make a lot of sense to introduce a coupled electronic currency that is mainly generated by Prestige: I call this currency Fluido. Fluido also has the nice property of being a truly continuous currency: Currency generation if continuous. Continuous transfers of Fluido are also possible. And it's easy to implement an optional basic income in Fluido.

    When it comes to implementing QP there are many different possible routes.
    • as centralized app (I want to create a first demo version in that way, so people can play around with it – in contrast to money, reputation economies are a rather unkown animal, so testing it in a centralized context first makes a lot of sense)
    • as extension for a centralized app
    • on its own blockchain (I have ideas for an adapted PoW scheme, but I feel there should be better possible approaches that take full use of the reputation system architecture)
    • on Ethereum
    The complexity of QP might be a problem for its implementation on Ethereum, so I would like to hear from you how you would judge the feasibility of this approach. Would QP use up too much gas to be viable? Would there be other possible problems?

    Initially I wanted to post this idea in a new thread, but I thought it would fit it nicely here. Do you think I should go ahead and make a dedicated QP thread anyway?
Sign In or Register to comment.