More actions
It can be difficult to imagine an economy without money. The idea of a “gift economy” sounds nice, but breaks down at any scale whatsoever due to people taking advantage of the system without providing anything in return (the “free-rider problem”). Money serves as a more-or-less objective measure (however flawed it may be) of contribution to an economy, and its exchange for goods and services serves as an imperfect but somewhat functional check against the free-rider problem. However, in theory, an economy without money may work if the free-rider problem could be mitigated via other means. One such means could be our decentralised system of ratings on predicates.
Assume that we have developed a open, active network of participants in society, who rate each other on a variety of criteria. Consider first a very simple predicate: “X is a net contributor to society.” Depending on the ratings of one’s trusted peers, the computed rating for this predicate will come back differently for every person who runs it. However, that rating can be used to inform an individual decision on whether to provide goods or services to that individual.
John, George, and the jar of pickles
Consider a scenario: George is a homesteader who cans his own pickles. John is a labourer who is down on his luck and has not had work for the better part of a month. He does not own land, nor does he have a stockpile of food, but he has routinely volunteered at a local animal shelter for years. He is hungry, and he has come to George’s door, explained the situation, and asked if George could spare him a jar of pickles.
Of course, if this is a tight-knit community, George may already know John, or at least he may have heard through the grapevine of John’s history of selfless activity in the community. But let’s assume that George not only doesn’t know John, he has no idea who he is. Without this knowledge, George now has a dilemma on his hands.
George’s Dilemma
George is not a very selfish person, in fact he enjoys helping those in need. However, he’s been around the block a few times. He knows that people who act selflessly are often taken advantage of, especially by strangers with sad stories. George has a limited stock of pickles preserved from the last growing season. He has more than enough to get him through to the next harvest, and he’s in no danger of starvation or even discomfort from giving away one jar of pickles. In fact, many of the jars of pickles that George has preserved have a pretty good chance of going to waste. He’s not a commercial farmer, just a cautious homesteader who prepares for the worst and then some.
Being an empathetic person with a taste for altruism (and pickles), George really wants to help John. But he knows that there are risks associated with doing so. John could be a con-man going door-to-door asking for something from everyone. Knowing that George is generous and responsive to a good sob story, he might start coming around asking for a handout on a regular basis. He might even tell his friends that George’s is a good place to trade an earful for a pickle.
Of course, in this story, we know that John is an upstanding citizen who is down on his luck. But George doesn’t know that. He weighs his options for a few seconds and decides that he can’t take the risk. George, like John, is an honest man and has never been a good liar. He mumbles an apology and something about not knowing if he has enough, wishes John well, and closes the door.
The other side of the tragedy of the commons
We have read about “the tragedy of the commons”: the idea that, given open access to an abundant but limited resource, a community will break down into a horde of selfish actors who each seek to maximise their advantage, thus depleting the resource and causing a hardship to the entire community, which could have been avoided. It’s not equivalent, but is adjacent to the free-rider problem, and it’s basically why a gift economy doesn’t work at scale. Setting aside actual exchange of value, we sometimes need to attach a cost to a good or service we would gladly give or perform for free, simply so that it doesn’t become attractive to those seeking handouts.
But there is another tragedy that comes from this: the tragedy we’ve just watched happen between John, a genuinely good person in need, and George, a genuinely good person who truly wanted to help John, but felt that he couldn’t. This tragedy starts with John walking away hungry, and fully manifests in George’s dumping out his excess pickles in the compost heap to make room for the new harvest next autumn.
While this particular instance is fictional, this phenomenon occurs constantly in the real world, and is a problem worth solving.
John and George in a better world, or, in the current world, with better tools
George’s reasoning behind denying John a jar of pickles is not unsound, and it’s not immoral. It is, however, tragic, since his first impulse would have been to help John. This was not a failure of George’s morality nor his rationality, simply his understandable ignorance of John’s nature and actions combined with a bleak but fair assessment of humanity as a whole.
Let’s suppose that George had the ability to instantly poll the community asking: “Is John a net contributor to society?” Having this information would change everything. George could rest assured that John was not a career beneficiary, and would therefore gladly help him. The pickles don’t go to waste, John gets to eat, and everyone is happy.
Problems with this suggestion
Such an ability doesn’t exist at the moment, of course. There are complications of such a system. Ignoring technical complications, to start, the idea of someone being a net contributor to society is very subjective. Imagine that a person is asked to rate John’s contributions to society, and they happen to dislike when humans prioritise the well-being of animals over that of humans. In that case, they are likely to rate John lower due to his history with the animal shelter.
Another problem is the question being asked. “Is John a net contributor to society” is arguably a decent question to ask when evaluating whether you should give an able-bodied man some pickles. It might be a less decent question to ask if John were disabled in a way that makes meaningful contributions difficult. Perhaps, then, a better question to ask might be, “Does John deserve help?”
But there’s another problem. “Does John deserve help” is maybe a better question in the pickle situation. But what if you’ve got some work you’d like to hire John to do? A better question to ask in this situation is “Does John do decent, honest work?”
A (theoretically) more effective solution
A technical definition of our system for rating predicates is beyond the scope of this writeup. Assume we have a system in which any yes-or-no question can be rated by any number of participants in the system, and client software which can, with the help of data from this network, provide a customised aggregate rating of any such question to the user.
It would then be possible for George to ask any combination of questions about John, and combine these answers to help him make his decision. The singular question could be replaced by many questions, perhaps:
- Is John being truthful about his identity?
- Does John volunteer at the animal shelter?
- Is John is an honest worker?
- Does John perform well in manual labour?
- Do I have enough pickles to last the season?
- Are adverse growing conditions expected?
This is getting complicated; too much so for a human to perform alone, perhaps (especially with a hungry man on one’s doorstep). A different approach would be to have a software application combine these answers for us. Really, the only question George has to answer is:
- Should I give John a jar of pickles?
In a world where everyone has this application installed, everyone is rating everyone else on everything, and the system is able to keep up, this would result in the creation of a parallel subjective economy. This would basically function as a technological check against the “pickle tragedy.” Everyone who genuinely needed help could seek it without fear of stigma. Those who did stigmatise genuine help-seeking would find themselves rated unfavourably when the shoe switched feet. In this way, we could organically both incentivise helping the needy, and disincentivise perpetrating the tragedy of the commons.
Collecting inputs to the ratings system
Of course, it is no use consulting a ratings network which has no data. The hypothetical situation with John and George took for granted that the network had sufficient data to answer the questions George was asking. But in order to have that data available, it is necessary to collect that data. How can this be done?
One way would be to prompt the entire network for a rating every time that rating was needed. This would theoretically work assuming a sufficient quorum of users were online and willing to provide a response. However, this is obviously not very practical for several reasons. Such a mechanism would be offensively noisy and users would quickly disable it or remove the application. As the network grew in size and usage, it would even become physically impossible to sustain. It would also be incredibly wasteful.
One improvement on this idea would be if the answers to certain questions were cached once the user had answered them once. This would at least stop the same question from being asked over and over. This isn’t enough, however, and it also introduces a new problem with changing information. Imagine that one day, a question comes through for rating: “Is Tony an honest worker?” You answer affirmatively, since he has done some roof work for you recently, charged a fair amount, and was generally pleasant. However, a few weeks later, your roof develops a leak again. After inspecting the issue you come to an unfortunate conclusion: Tony’s “fix” was extremely temporary. You try calling his number, but the line is dead. You’ve been had.
It is clearly important for users not only to provide as many ratings as possible, but to update those ratings as new information becomes available to them. One way to handle this would be to prepare a “ratings review” for the user to complete every so often. This review could provide a manageable selection of prior ratings, and perhaps some popularly-requested predicates on which that user has not yet given a rating, and ask the user to (re)rate those items.
This sort of automated, asynchronous request for ratings could form just one component of input to the system. There are better ways to acquire certain data. For example, certain ratings opportunities related to transactions between individuals may be intelligently inferred by the application. In the case of John and George, for example, George may choose to provide an affirmative rating on a predicate like “John has recently asked for a favour,” or perhaps he may provide a specific transaction record to the network. John may be notified of this by his application, and be asked to affirm the transaction record or provide a rating on George’s generosity.
The application could also prompt for ratings in this way based on other information. If the correct permissions are enabled and consent is given by the user, the user’s location or proximity to other devices could be used to inform ratings prompts. For example, even if neither John nor George immediately thought to record the transaction, they could be prompted to answer questions about their interaction later on, based on the fact that their devices were in close proximity for a certain time.