How do you distinguish precommittments from simple contracts?
If you are standing in the market selling apples for dollar a pound, have you precommitted to anything?
Generally speaking, precommittments are expensive because you pay with optionality, the ability to make a choice later. There must be a good reason to precommit, something other than "wouldn't it be generally useful".
Precommitments are more general, since they don't require more than one party, but they are very similar.
Currently, contracts are usually enforced by the government, and there are limits to what can be included in a contract, and the legality of the contract can be disputed.
Binding precommitments would be useful for enabling cooperation in inefficient games: http://lesswrong.com/lw/nv3/inefficient_games/
Why should they? If you want to have votes decide over specific issues instead of deciding about candidates during a referendum on the issue is a lot more straightforward.
Why should they?
To bring what voters think they're getting from each candidate and what voters will actually get from each candidate closer together. Voters are still "deciding about candidates" rather than deciding each issue separately, but this way they would have better information about the candidates.
Infamous recent example: The UK just held a referendum about whether it should leave the European Union. The "Leave" campaign repeated, over and over again, that the UK sends £350M to the EU every week and that leaving the EU would enable that money to be spent on the National Health Service (widely considered by the UK population to be awesome apart from being underfunded). The only trouble is that (1) the UK doesn't send anything like £350M/week to the EU, (2) even if that figure were right it will almost certainly be dwarfed by other economic effects of leaving the EU, and (3) if the UK did save that much money by leaving the EU, none of the leaders of the Leave campaign would actually give the money to the NHS anyway. So as soon as the result of the referendum was announced (like, within about an hour) the Leave people started saying: oh, no, of course we aren't going to do that, we never said we would or if we did we shouldn't and OH HEY LOOK OVER THERE AN IMMIGRANT.
Now, actually in this case it's not clear how a law of this sort could work. Who would be on the hook if Leave won (as they did) and then an extra £350M/year didn't go to the NHS (as it won't)? The senior figures in the Leave campaign don't actually have the power to control the budget. The people who do control the budget (the prime minister and the Chancellor of the Exchequer) were on the Remain side. And most of the actual propaganda was written and distributed by other people.
I suspect the main result of a law like this might just be to increase the amount of indirection between candidates and the promises made during their campaigns. "Oh no, I never promised that. That was done by the Campaign to Re-Elect Lord Voldemort. It's nice that they support me but I don't control what their advertisements say." Some categories of campaigning can be protected from this by requirements to have "I am Lord Voldemort and I approve this message" attached, but I think actual political propaganda is more and more happening on social media and being done not by candidates and their closely-tied supporters but by Just Plain Folks with a TwitFaceChatGram account.
The other downside would be that when candidates' promises and actions don't match, often the actions are better than the promises. Suppose Donald Trump is elected president. Do you really want him to put up an enormous reinforced concrete wall all along the US/Mexico border? If he does, do you really want him to try to strongarm Mexico into paying? A law imposing severe penalties for breaking campaign promises doesn't incentivize better promises so much as it incentivizes following through on promises even if they're stupid ones.
I do see the appeal of a law like this, but I think it would be really hard to make it actually do more good than harm.
To bring what voters think they're getting from each candidate and what voters will actually get from each candidate closer together.
As far as I understand you argue that it's not of the job of the voters to hold politicians accountable and make their voting decisions by evaluating the track record of politicians but instead some other entity should be tasked with holding politicians accountable.
Infamous recent example: The UK just held a referendum about whether it should leave the European Union.
Is that supposed to be an example that suggests why holding a referndum is a bad idea?
As far as I understand you argue [...]
No, I am not arguing that it isn't the job of the voters to hold politicians accountable. That diverges from what I am actually saying in two ways.
The proposal in question (which, I repeat, is not mine) does, however, argue that as well as the usual means of keeping politicians accountable, there should be a further means of doing so. Voting is a very blunt instrument and it's not obviously wrong to augment it.
Is that supposed to be an example that suggests why holding a referendum is a bad idea?
No, it's supposed to be an illustration of a situation in which politician's pre-election campaigning and their post-election behaviourshowed a particularly stark mismatch. As it happens I do think the UK's referendum was a silly idea badly executed, but that wasn't at all why I was bringing it up here. (That would have been a total non sequitur.)
I am not proposing, nor do I think anyone here is proposing, to abolish elections, whereby the voters hold politicians accountable.
If voters have the impression that they should choose candidates based on the promises those candidates made for the future they will put less effort into rewarding past actions and holding politicians accountable for past actions.
I want politicians you listen to expert advice and changing circumstances and are willing to change their mind about what has to be done. I don't want that our political system turns into a copy of the soviet system that replaces 5-year plans with 4-year plans. Advocating that an election for a president should be understood as agreeing to a 4-year waterfall plan is in my mind likely to damage our democracy from functioning properly.
Yes, I agree; that's one of the reasons why "often the actions are better than the promises". Though candidates' promises are generally nothing as specific as 4-year waterfall plans, and I would expect them to get less specific if a law of this sort were introduced.
In Germany there was a campaign promise of the SPD at the last election to not change the current law of accepting refugees.
The SPD broke that promise when the amount of refugee seekers rose drastically. I think that fact that they did so, shows that they react intelligently to changing circumstances. I wouldn't want to have politicians looked into 4-year promises that prevent them from acting dynamically.
Binding promises also make consensus finding harder and could even forbid it when the promises made before the election aren't compatible.
We already have legal contracts to do this. If I make a website and sell a product I however people to cooperate. They can make a contract with me and then I am precommitted to deliever them the product they paid for.
Ethereum contracts are also limited in what they can include. It's not qualitative different in that regard.
require a government to enforce them.
We are lucky and have governments in the western world that enforce our contracts and thus provide economic properity.
open-source prisoner's dilemma
I believe the GNU GPL was made to address this.
It seems like we are moving in this direction, with things like Etherium that enable smart contracts.
Does anyone have proof that Etherium is secure? There's also the issue of giving whomever runs Etherium complete authority over those 'smart contracts', and that could easily turn into 'pay me to make the contract even smarter'.
Technology should enable us to enforce more real-world precommitments, since we'll be able to more easily monitor and make public our private data.
People are going to adapt. And I see no reason why would anybody share particularly private stuff with everyone.
And then there's the part where things look so awesome they can easily become bad: I can imagine someone being blackmailed into one of those contracts. And plenty of other, 'welcome to the void' kind of stuff.* Where's Voldie when you need him?
People will be incentivized to share private things if robust public precommitments become available, because we all stand to benefit from more information. Because of human nature, we might settle on some agreement where some information is private, or differentially private, and/or where private information is only accessed via secure computation to determine things relevant to the public interest.
We have precommitments already. It's just that every time someone follows through on one, people at LW are eager to jump on them for being irrational because they obviously made the choice that produces less of what they want than some alternative choice. But emotional reactions that predictably lead to "irrational" behavior are forms of precommitment.
Of course this doesn't lead to arbitrary precommitments.
Does anyone have proof that Etherium is secure?
Define "secure". And, naturally, Etherium contracts live on the blockchain, so there is no one who "runs" Etherium in the same sense that there is no one who runs Bitcoin. But, of course, persuade a sufficiently large part of the community and you can have anything you want -- see the DAO mess and the consequent hard fork.
The ability to make arbitrary public binding precommitments seems like a powerful tool for solving coordination problems.
We'd like to be able to commit to cooperating with anyone who will cooperate with us, as in the open-source prisoner's dilemma (although this simple case is still an open problem, AFAIK). But we should be able to do this piece-meal.
It seems like we are moving in this direction, with things like Etherium that enable smart contracts. Technology should enable us to enforce more real-world precommitments, since we'll be able to more easily monitor and make public our private data.
Optimistically, I think this could allow us to solve coordination issues robustly enough to have a very low probability of any individual actor making an unsafe AI. This would require a lot of people to make the right kind of precommitments.
I'm guesing there are a lot of potential downsides and ways it could go wrong, which y'all might want to point out.