The Anonymous, Recursive Suggestion Box
Good discussion with Ross today, resulting in one nice, concrete idea.
Consider the problem of suggesting policy improvements to the government. In particular, let’s imagine someone has a specific, detailed policy change related to health care, financial regulation, etc. Presumably, the people who know the most about these industries are (or were) in the industries themselves, so you could argue that they can’t be trusted to propose ideas that aren’t just self-serving. Maybe it’s possible for someone to build a reputation of trustworthiness, but that’s hard and would ideally be unrelated to the actual ideas proposed. Instead of relying on reputation, we’ll remove the issue entirely by making the suggestion box anonymous.
Now we have an anonymous suggestion box on a website. People go to it and propose ideas. There are a few good ideas, and a vast amount of bad, malicious, and nonsense ideas (including spam). Eliminating the spam is easy (I have a single, completely public email address and get roughly one spam message per day, from which I conclude that the spam issue is solved). In order to eliminate the bad or malicious ideas, we need to be able to judge their correctness in a logical manner. For this, we rely on distributed intelligence: other people are allowed to judge whether each idea is good or bad. To get loaded words out of the picture, let’s replace “good” and “bad” with the words “true” and “false”. “Ideas” become propositions of the form “Implementing this idea would be good” (yes, “good” is still there, but keep reading).
Let’s assuming voting isn’t a completely reliable system for determine the truth or falsity of ideas (otherwise, we’re done). Therefore, some true propositions will get a lot of false votes, and vice versa. To solve this, we allow people to propose arguments for or against each proposition. These have the form of some statement, like “That proposition is false because the author is a moron”, together with a more details argument for why the statement is true. Now we let people vote on two more things:
- Whether the truth of the statement would imply that the original proposition is true or false.
- Whether the argument for the truth of the statement itself is sound.
If we get enough votes in favor of both (1) and (2), we conclude that the original idea is true (or false), and discount the votes for or against the original idea.
This is the key part, so I’ll restate it. If we have propositions , , and , then enough votes for both and override any votes against . You can’t kill a good idea unless you can the arguments for it as well.
Now we have to get recursive. What if gets a lot of votes, but is actually wrong? Then you let people propose arguments for the falsity of , and so on. What if there are two competing arguments which appear to contradict? Then you let people propose arguments about why there isn’t a contradiction? There are a lot of logical issues to deal with, but people can post arbitrary arguments written in normal human languages and we have the full power of human intelligence to judge them, so we’re not limited by artificial logical restrictions. This isn’t a formal proof system.
Unfortunately, we are limited by what happens along the full recursive tree. If people lie about the propositions all the way down, and manage to flood away all the counter arguments, the system will fail. However, this is basically a problem of spam, and can be solved in the usual way. If you detect that someone is consistently voting opposite the correct answer, you flag them as malicious and discount their votes. This rule is circular, but that’s what probabilistic analysis is for: we take all the data and compute the most likely assignment of truth values to propositions and spam flags to people. There’s some threshold of validity that you need to achieve in order to such a solver to converge to the correct answer, but that level of trust is often quite low due to network effects and self-reinforcement. In other words, contradictions don’t fit together.
Since this is a website, we have to identify whether the “users” are actually people. We could do this conventionally with a system like ReCAPTCHA, but since we’re in recursive mode it’s much cooler to instead ask users to judge the correctness of randomly selected propositions. If you want to vote on whether a proposition is true or false, or propose a new proposition, you need to spent a little time judging the ideas of others. If someone comes up with a way to trick this system by writing a program that can judge the truth or falsity or arbitrary English propositions, this discussion may be obsolete (thanks to Ross for this particular bit of reasoning).
Other issues probably abound, but they can be fixed by allowing people to suggest improvements to the system. If deemed reasonable, these ideas can be implemented and tested in parallel with the existing system, resulting in a potentially large number of competing systems for determining truth values from the same data set. The data set itself could probably be made freely available (under a suitable license), so that others could build competing systems.
I don’t think this system would be all that difficult to implement. Thanks to the previous paragraph, if it reached a sufficient level of quality it would start to improve itself. Maybe that would even get scary.
Of course, if we apply this to a realm like politics, the truth or falsity of various statements will be very controversial, and different people will have legitimately different opinions. This can be solved by adding side conditions to the statements, like “If you believe in flat tax systems, we should do this” or “If you believe that health care is a basic human right, we should do that.” More importantly, however, there is a vast range of ideas that any rational person should agree to. Statements like “the proposed health care bills do not include death panels”, and “given an otherwise equivalent choice between taxing a public good and a public evil, we should tax the latter.” I think it’s fair to say that the U.S. would be better off if we could agree on the statements that don’t need side conditions.
Note: I’ve done zero checking to see if this has been proposed or implemented before (this discussion happen just now), so I’m curious if anyone knows related references or links.
Another note: presumably this would be set up as a nonprofit supported by donations of some kind. If this system actually existed, I would probably be willing to donate at least $1000.
Tags: distributed system, suggestion box, voting
August 16th, 2009 at 8:23 pm
This is a cool idea unto itself… I have some concerns about the philosophy it embodies: namely the notion of central planning.
Put aside the hyperbole of the following passage from Hayek and focus on his key claim:
Here is another problem: I would fail your “death panels” test–because I see quite a bit of subtly to the question. Its all about what the meaning of “is” is. Reform plan defenders have redefined the term “death panel” to be something that they could definitively refute as false. Suddenly it became a question of life counseling and assisted suicide. Huh, say what. Yes I agree that form of the claim of death panels is nonsense, but it is also a strawman.
Ann Althouse had a very good discussion of the backstory: http://althouse.blogspot.com/2009/08/did-sarah-palin-say-obamas-death-panel.html
“Death Panel” is the sense it was introduced is a very charged phrase that captures a certain element of the consequences of a public-plan. Consequences that are not unique to public-plan–Private insurers have death panels too in that sense, but I refer you to my earlier quote from Hayek.
Political discourse is dishonest.
August 17th, 2009 at 2:53 am
Thanks for the comments. I’ll try to reply as if I was using the system described.
1. Your quote from Hayek is irrelevant since the suggestion box does not embody a central planning philosophy (for the record, I agree with the quote). I’ll give three arguments for this:
a. It isn’t central planning because anyone is free to contribute ideas and arguments, and the system is fundamentally based on voting by individuals.
b. If you’re concerned about the power of the website admins, the problem is solved by maknig the data set free and allowing other people to set up competing sites, sharing data or not as they wish.
c. If you’re concerned that the ideas are coming from “insiders”, this concern is groundless because the suggestions are anonymous. Even if they come from insiders, they will only be labeled as true if other people understand the arguments for them and agree.
2. Of course the “death panel” thing is a strawman. This is in accord with my system since strawman arguments would hopefully be quickly eliminated. Your link is irrelevant because we both agree the subject is a strawman.
3. In the same paragraph, you appear to say that there’s real subtlety behind it. I.e., that it isn’t a strawman. However, your argument appears to hinge on “what the meaning of ‘is’ is”. I’m fairly confident I could get an overwhelming majority of people to vote for the notion that “what the meaning of ‘is’ is” is devote of meaningful content. If you claim there’s a real issue driving this conflict (which there is), you would be free to propose a modified proposition stripped of the loaded “death panels” phase.
4. You claim that “Political discourse is dishonest.” This is false as stated; a correct version is “Some but not all political discourse is dishonest.” The corrected version is evidence in favor of the suggestion box, since it could be a useful tool for distinguishing which is which.
August 17th, 2009 at 3:07 am
Oops, I forgot the most important reason this isn’t central planning:
The suggestion box wouldn’t have any legal force. People would propose ideas or facts, the hopefully the system would be a useful tool for checking them. If it had sufficient credibility, you could send people links to it so that they could read the arguments, or politicians could use it as a source of ideas, but I’m certainly not advocating that we’d immediately implement all of the ideas that end up labeled true.
August 17th, 2009 at 8:38 pm
1) Your three arguments do not enclose all the ways in which the proposal may embody the philosophy of central planning. In particular, I believe it does so by including the premise of a unitary result as a sin qua non.
Further responses to your particular counterarguments: 1a) The character of central planning is unchanged, even if the plan is ratified democratically. There would nonetheless be one result prescribed onto everyone as truth.
1bc) I do not object. I am not concerned about the administrators. I do not believe that accepting this point is otherwise dispositive.
1d) The lack of legal force means that it is not central planning itself. Embodying a philosophy is different from this.
2) I said that the common rendition of ‘Death Panel’ is a strawman established by proponents of “the” bill to make the claim ridiculous. I agree that under their terms, there are no ‘Death Panels’ in the bills under discussion.
3) We agree that “death panel” is not a phrase with intrinsic meaning. Therefore it can be ascribed a meaning with some latitude. Thus the term “is” has uncertain implications. The original issue raised and labeled “death panel” is an implied consequence of a government panel deciding which treatments merit coverage. Private insurers do the same, but multitude of decision making bodies makes any given determination diffuse–which is the concern raised by Hayek.
4) The statement is true without additional qualifiers given the rules of standard English. For instance, the statement ‘the sky is blue’ would be considered correct under common definitions; but it is equivalent to saying ‘the sky is usually blue’ or ‘the sky can be blue’ rather than saying ‘the sky is always blue’. A challenge for the suggestion system would be that English is not a formal system–it easily allows ambiguous claims even when ambiguity is not the intention.
August 18th, 2009 at 2:13 am
If someone comes up with a way to trick this system by writing a program that can judge the truth or falsity or arbitrary English propositions, this discussion may be obsolete
Perhaps not a program with an artificial intelligence guiding those decisions, but I could see a program that farmed it out to something like mturk for uninvolved humans to make those judgments.
August 18th, 2009 at 5:31 am
Jon:
1. I do not have the premise of a unitary result in all cases. However, I do strongly believe that some statements do have unitary results, and that this class of statements is large enough to be useful. For statements which have no such result, the system would not be able to conclude one way or the other. In fact, it would probably conclude with a unitary result of “the original statement has no unitary result.”
Here’s an example of a nontrivial statement which I believe has a unitary result: “Taken as a single event, extending the term of copyright for an existing work does not encourage the production of new works.”
Thus, if we assume that the purpose of copyright is to encourage creative work (an assumption with no unitary result), and we believe that public policy decisions should be made based on the merit of the decision alone rather than on the message it sends about possible future legislation (again, no unitary result here), then we should not extend the term of existing copyrights.
In addition, I claim that the previous paragraph taken as a whole also has a unitary result; the assumptions should imply the conclusion regardless of your political views. I may also be missing some necessary side conditions, in which case other people would point them out and correct me.
1a. For better or worse, we have a extremely large body of laws that apply to everyone equally. If you’re equating this with central planning regardless of how the laws are ratified, then I agree then my system could be considered to have a central planning philosophy. However, I do not agree with this sweeping definition of central planning.
1d. Agreed.
2. Agreed.
3. Actually, I think the original statement in a bill that gave rise to the death panel phrase as a requirement that if people want end-of-life counseling, insurers would have to pay for it. The issue you raise is a real one, and could even be considered to apply to the “death panel” bit since it’s government deciding that counseling should be covered, but usually people cite the “death panel” claim as being about government deciding not to cover certain treatments.
The way this should work is that your real concerns should be split off from the death panel issue entirely, so that future people judging their merit wouldn’t have to wade through the garbage first.
4. This is a huge issue for a formal proof system, but I don’t think it’s a problem in the context of a human-based system. In the context of your argument, it seemed to me that your statement was intending to argue that political discourse is fundamentally dishonest, and therefore that attempting to improve it is futile. I apologize if I was misinterpreting the goal of your statement. In the real system, your statement would be explicitly attached to another making it clear how you were trying to use it, so I would be less likely to make this mistake.
The way I imagine this working is that as people set up arguments, the system might well infer a completely incorrect result. Others would notice this, and the system would give them a rough diagram of the important steps in the argument, and allow them to correct the holes.
August 18th, 2009 at 5:44 am
AzureLunatic:
This is certainly a concern, but I’m optimistic that it could be fixed. The key is that the results of the system are not immediately implemented. If someone implements an organized attack like the one you mention, I would expect it to initially work (not least because they might have the source code and be able to predict the result). If the attack goes unnoticed, it doesn’t matter since presumably that means people don’t disagree with what it accomplished. If it is noticed, the system can presumably be extended to recognize and reject the attackers. If such an extension is controversial, the system can be forked to maintain the unmodified system for comparison.