In "Liars and Outliers", Bruce Schneier talks about reputation as a tool that society uses to force its members to behave well. In small societies, reputation by itself is usually enough, but the problem is that it doesn't scale because people can ordinarily only keep track of about a hundred to a hundred and fifty people (Dunbar's number).
I'm of the opinion that, since Dunbar's number is a purely biological limitation, there's no reason to think that computer-assisted reputation can't scale to much larger groups of people. I think Cory Doctorow is right to be worried that an authoritarian government could use reputation as a tool of coercion (and this is why we should worry about things like government collection of phone metadata). However, I think it's also worth considering that reputation could be used as a tool by the people to obtain some accountability from the rich and powerful, and to help distinguish grass-roots support from astroturf. (I went so far as to create a site, http://polink.org, as a sort of collaborative graph database of how powerful people and organizations are connected to each other. It's one of those side projects I feel like I should get back to and work on one of these days.)
One of the reasons I think computer-assisted reputation hasn't taken off is that it's actually kind of hard to implement a good reputation engine, or even agree on what sort of properties a good reputation system ought to have. A lot of awful reputation systems have taught the technology industry to have very low expectations, which is a shame because I think they can be very powerful if implemented well and used correctly.
> I think Cory Doctorow is right to be worried that an authoritarian government could use reputation as a tool of coercion
Not just "could", China is already doing it. It sounds awful
But Peeple is a modest effort compared to ‘‘Citizen
Scores,’’ the for-now-voluntary service run by the Chinese
government in partnership with Tencent (a huge social media
and games company) and Alibaba (China’s answer to Amazon).
Your citizen score is visible to everyone the government
wants – buying socially approved items, undertaking
approved leisure activities, adhering to rules and
regulations, and socializing with other high-score
individuals. Of course, not doing these things makes your
score go down. Just being friends with low-scoring
individuals drags your own score down, creating a powerful
incentive to conform.
Mandatory Citizen Scores are being phased in over the next decade [...]
..tangentially, why even bother with reputations? The big issue with money based economy is that money is all cleansing. All the shitty things you did to get your money can be forgotten and your money used to buy yourself into a position of power, prestige, or even good standing.
If we could tag money's history of exchange we could maybe avoid the both the issues of reputations (big if obviously). Individual transactions could be judged by the prior history of exchanges. Though this could be used as a proxy for reputation or to construct a reputation, the reputation is not required to complete the transaction as a seller. The seller only needs judge the transaction on the history of the money offered itself.
Ideally this would create a situation where 'clean' currency would progressively become more valuable than 'dirty' currency, encouraging clean wealth creation.
I think it's pretty important for money to be fungible. Your idea is intended to solve a very real problem, but I think your proposed solution is worse than the problem.
I agree regarding the importance of fungability, but not all the way. If a majority of reputation engines only advised rejection of payments as offered by the bottom 1% of ranked teams, then 99% of teams would still experience the fungability expectation.
Otherwise, if the reputation engines are too strict, no one would trusts the currency. Too loose, and there is no effective deterence to unsustainable activity. There's an ideal balance to the targeted overall rejection rate among all teams, most likely erring on the side of being too loose (0.5% rejection?), filtering out only the worst of the worst. This targeting to a metric value reminds me of fed actions to indirectly influence the economy, such as with setting interest rates.
I like the idea of being able to reliably track how money was earned or used at the time that it is being offered as payment. However, it seems (to me, at least) that there is no practical way to do this for physical currency notes, for various reasons.
Fortunately, for digital currency there are easier ways to implement such tracking ideas. One approach is for a team to use its planned budgets directly as digital currency. When two teams transact, corresponding amounts of expense and revenue budgets get cancelled, so there is no transfer of currency between teams. Thus, each payment offer is always traceable to the reputation of the team who issued the currency (as budgets).
Illustrated explanations, more details, and a working prototype is available at: https://tatag.cc/ui/. (Disclaimer: I'm the developer of the linked site.)
Do not accept payments from teams who have budget balances that are way beyond typical. The Tatag platform provides accurate and current summary statistics about the paying team, so a reputation engine could factor those metrics to evaluate the concerns that are important to a payment recipient.
As an aside, traditional systems do not prevent similar situations from happening. You have ultra-wealthy people that are each effectively holding an endless supply of money. Should we not be asking that same question whenever and wherever we see it happen today?
>this is why we should worry about things like government collection of phone metadata). However, I think it's also worth considering that reputation could be used as a tool by the people to obtain some accountability from the rich and powerful
I find myself in agreement with much of what you say, but the above argument always makes me pause in these discussions. Essentially, the argument is: "on the one hand, the rich can use this tech to control the poor. That's bad. On the other, the poor can control the rich. That's good."
Yeah, well, which one are you? Where do you draw that line between who is "rich" and "poor"? Why is one side coercive, while the other is obtaining accountability?
We're all, here, the poor and oppressed, of course, only willing to use our new-found power for good.
It's about finding the right balance. If members of society weren't accountable at all to the rest of society, we would have absolute anarchy. If we had perfect compliance, we would have a totalitarian state. Schneier makes this point very well in his book - it's good that society can force cooperation most of the time, but it's also good that it's not able to force cooperation from everyone, all the time.
Reputation is a powerful (and rather dangerous) tool and software-assisted reputation calculation is something that's going to exist and can't be un-invented. Therefore, I think it would be best if that tool were accessible to all sides of society so it isn't exclusively applied by the powerful against the weak. I don't mean to imply that any use of reputation systems by the powerful is bad and any use by the less-powerful is inherently good, I just think that abuse is less likely if those wielding the tool are fully aware that the same tool can and will be used against them if they get enough people mad.
The approach I prefer is to let each market participant or team decide which reputation engine they want to use when evaluating a payment offer. So there does not have to be a global consensus on what reputation engine everyone should use.
I have prototyped a reputation currency platform [1] where third-party 'advisor' applications offer real-time advisory on whether a payment recipient should approve or reject a payment offer. Data science/big data techniques would help improve and adapt such reputation engines over time, much like web search engines improving incrementally over time.
I like the idea of making reputation data available to everyone and let them run their own calculations however they want. Unfortunately (or fortunately if you're wary of the privacy implications), most online applications that have some notion of reputation don't share reputation data don't allow you to just grab the whole database.
A team's reputation does not need to involve divulging personal information about its members. Although, if a team is small enough, there's a good chance that generalized transactions reported between teams may still be matched to a particular team member.
The bigger picture is that a team in the proposed system would not have to constantly worry about funding, since it could use its budgets directly as currency as long as the team maintains its reputation. I think that for most participants that is more than good enough trade-off to take on small privacy risk. In other words, would a team rather worry about funding or privacy, and with good platform design those concerns might not even be mutually exclusive in most cases.
With regards to not having access to reputation data from online applications, there are work-arounds. For example, tweets, blog posts, or product reviews about a team could be crawled for sentiment detection and used in reputation metric calculations. These work-arounds might seem difficult to do, but so was search engine technology in the early days of the web.
I wonder if there is another aspect in play, the mutability of identities.
In small, local communities, you have one thing you can never get rid off, your face.
But online there is no face, nothing to tie your claimed identity back to your biological self.
While pondering this i reminded myself of a mmorpg that may illustrate this.
It originated in South Korea, and there the player account is tied to the players national identity number. So once banned, one is banned for life.
But when it was exported to the states, anyone could sign up multiple times. End result was that various mechanics that originally was meant for the community to police itself, was used to harass other players.
That is an important characteristic of online communities: someone with a bad reputation can usually (depending on the application) throw away their old account and start over with a new identity; therefore, having no reputation is more-or-less indistinguishable from having a negative reputation. That's not really a problem per se, it's just a characteristic of the problem domain to consider when implementing a reputation systems (and in particular, in deciding how much leeway to impact the system to give to new users).
I'm of the opinion that, since Dunbar's number is a purely biological limitation, there's no reason to think that computer-assisted reputation can't scale to much larger groups of people. I think Cory Doctorow is right to be worried that an authoritarian government could use reputation as a tool of coercion (and this is why we should worry about things like government collection of phone metadata). However, I think it's also worth considering that reputation could be used as a tool by the people to obtain some accountability from the rich and powerful, and to help distinguish grass-roots support from astroturf. (I went so far as to create a site, http://polink.org, as a sort of collaborative graph database of how powerful people and organizations are connected to each other. It's one of those side projects I feel like I should get back to and work on one of these days.)
One of the reasons I think computer-assisted reputation hasn't taken off is that it's actually kind of hard to implement a good reputation engine, or even agree on what sort of properties a good reputation system ought to have. A lot of awful reputation systems have taught the technology industry to have very low expectations, which is a shame because I think they can be very powerful if implemented well and used correctly.