Forecasting on LessWrong: I’ve been thinking for quite a while about somehow integrating forecasts and prediction-market like stuff into LessWrong. Arbital has these small forecasting boxes that look like this:
I generally liked these, and think they provided a good amount of value to the platform. I think our implementation would probably take up less space, but the broad gist of Arbital’s implementation seems like a good first pass.
I do also have some concerns about forecasting and prediction markets. In particular I have a sense that philosophical and mathematical progress only rarely benefits from attaching concrete probabilities to things, and more works via mathematical proof and trying to achieve very high confidence on some simple claims by ruling out all other interpretations as obviously contradictory. I am worried that emphasizing probability much more on the site would make making progress on those kinds of issues harder.
I also think a lot of intellectual progress is primarily ontological, and given my experience with existing forecasting platforms and Zvi’s sequence on prediction markets, they are not very good at resolving ontological confusions and often seem to actively hinder them by causing lots of sunk-costs into easy-to-operationalize ontologies that tend to dominate the platforms.
And then there is the question of whether we want to go full-on internal prediction market and have active markets that are traded in some kind of virtual currency that people actually care about. I think there is a lot of value in that direction, but it’s obviously also a lot of engineering effort that isn’t obviously worth it. It seems likely better to wait until a project like foretold.io has reached maturity and then see whether we can integrate it into LessWrong somehow.
This feature is important to me. It might turn out to be a dud, but I would be excited to experiment with it. If it was available in a way that was portable to other websites as well, that would be even more exciting to me (e.g. I could do this in my base blog).
Note that this feature can be used for more than forecasting. One key use case on Arbital was to see who was willing to endorse or disagree with, to what extent, various claims relevant to the post. That seemed very useful.
I don’t think having internal betting markets is going to add enough value to justify the costs involved. Especially since it both can’t be real money (for legal reasons, etc) and can’t not be real money if it’s going to do what it needs to do.
There are some external platforms that one could integrate with, here is one that is run by some EA-adjacent people: https://www.empiricast.com/
I am currently confused about whether using an external service is a good idea. In some sense it makes things mode modular, but it also limits the UI design-space a lot and lengthens the feedback loop. I think I am currently tending towards rolling our own solution and maybe allowing others to integrate it into their site.
One small thing you could do is to have probability tools be collapsed by default on any AIAF posts (and maybe even on the LW versions of AIAF posts).
Also, maybe someone should write a blog post that’s a canonical reference for ‘the relevant risks of using probabilities that haven’t already been written up’, in advance of the feature being released. Then you could just link to that a bunch. (Maybe even include it in the post that explains how the probability tools work, and/or link to that post from all instances of the probability tool.)
Another idea: Arbital had a mix of (1) ‘specialized pages that just include a single probability poll and nothing else’; (2) ‘pages that are mainly just about listing a ton of probability polls’; and (3) ‘pages that have a bunch of other content but incidentally include some probability polls’.
If probability polls on LW mostly looked like 1 and 2 rather than 3, then that might make it easier to distinguish the parts of LW that should be very probability-focused from the parts that shouldn’t. I.e., you could avoid adding Arbital’s feature for easily embedding probability polls in arbitrary posts (and/or arbitrary comments), and instead treat this more as a distinct kind of page, like ‘Questions’.
You could still link to the ‘Probability’ pages prominently in your post, but the reduced prominence and site support might cause there to be less social pressure for people to avoid writing/posting things out of fears like ‘if I don’t provide probability assignments for all my claims in this blog post, or don’t add a probability poll about something at the end, will I be seen as a Bad Rationalist?’
Also, if you do something Arbital-like, I’d find it valuable if the interface encourages people to keep updating their probabilities later as they change. E.g., some (preferably optional) way of tracking how your view has changed over time. Probably also make it easy for people to re-vote without checking (and getting anchored by) their old probability assignment, for people who want that.
Note that Paul Christiano warns against encouraging sluggish updating by massively publicising people’s updates and judging them on it. Not sure what implementation details this suggests yet, but I do want to think about it.
Yeah, strong upvote to this point. Having an Arbital-style system where people’s probabilities aren’t prominently timestamped might be the worst of both worlds, though, since it discourages updating and makes it look like most people never do it.
I have an intuition that something socially good might be achieved by seeing high-status rationalists treat ass numbers as ass numbers, brazenly assign wildly different probabilities to the same proposition week-by-week, etc., especially if this is a casual and incidental thing rather than being the focus of any blog posts or comments. This might work better, though, if the earlier probabilities vanish by default and only show up again if the user decides to highlight them.
(Also, if a user repeatedly abuses this feature to look a lot more accurate than they really were, this warrants mod intervention IMO.)
Forecasting on LessWrong: I’ve been thinking for quite a while about somehow integrating forecasts and prediction-market like stuff into LessWrong. Arbital has these small forecasting boxes that look like this:
I generally liked these, and think they provided a good amount of value to the platform. I think our implementation would probably take up less space, but the broad gist of Arbital’s implementation seems like a good first pass.
I do also have some concerns about forecasting and prediction markets. In particular I have a sense that philosophical and mathematical progress only rarely benefits from attaching concrete probabilities to things, and more works via mathematical proof and trying to achieve very high confidence on some simple claims by ruling out all other interpretations as obviously contradictory. I am worried that emphasizing probability much more on the site would make making progress on those kinds of issues harder.
I also think a lot of intellectual progress is primarily ontological, and given my experience with existing forecasting platforms and Zvi’s sequence on prediction markets, they are not very good at resolving ontological confusions and often seem to actively hinder them by causing lots of sunk-costs into easy-to-operationalize ontologies that tend to dominate the platforms.
And then there is the question of whether we want to go full-on internal prediction market and have active markets that are traded in some kind of virtual currency that people actually care about. I think there is a lot of value in that direction, but it’s obviously also a lot of engineering effort that isn’t obviously worth it. It seems likely better to wait until a project like foretold.io has reached maturity and then see whether we can integrate it into LessWrong somehow.
This feature is important to me. It might turn out to be a dud, but I would be excited to experiment with it. If it was available in a way that was portable to other websites as well, that would be even more exciting to me (e.g. I could do this in my base blog).
Note that this feature can be used for more than forecasting. One key use case on Arbital was to see who was willing to endorse or disagree with, to what extent, various claims relevant to the post. That seemed very useful.
I don’t think having internal betting markets is going to add enough value to justify the costs involved. Especially since it both can’t be real money (for legal reasons, etc) and can’t not be real money if it’s going to do what it needs to do.
There are some external platforms that one could integrate with, here is one that is run by some EA-adjacent people: https://www.empiricast.com/
I am currently confused about whether using an external service is a good idea. In some sense it makes things mode modular, but it also limits the UI design-space a lot and lengthens the feedback loop. I think I am currently tending towards rolling our own solution and maybe allowing others to integrate it into their site.
One small thing you could do is to have probability tools be collapsed by default on any AIAF posts (and maybe even on the LW versions of AIAF posts).
Also, maybe someone should write a blog post that’s a canonical reference for ‘the relevant risks of using probabilities that haven’t already been written up’, in advance of the feature being released. Then you could just link to that a bunch. (Maybe even include it in the post that explains how the probability tools work, and/or link to that post from all instances of the probability tool.)
Another idea: Arbital had a mix of (1) ‘specialized pages that just include a single probability poll and nothing else’; (2) ‘pages that are mainly just about listing a ton of probability polls’; and (3) ‘pages that have a bunch of other content but incidentally include some probability polls’.
If probability polls on LW mostly looked like 1 and 2 rather than 3, then that might make it easier to distinguish the parts of LW that should be very probability-focused from the parts that shouldn’t. I.e., you could avoid adding Arbital’s feature for easily embedding probability polls in arbitrary posts (and/or arbitrary comments), and instead treat this more as a distinct kind of page, like ‘Questions’.
You could still link to the ‘Probability’ pages prominently in your post, but the reduced prominence and site support might cause there to be less social pressure for people to avoid writing/posting things out of fears like ‘if I don’t provide probability assignments for all my claims in this blog post, or don’t add a probability poll about something at the end, will I be seen as a Bad Rationalist?’
Also, if you do something Arbital-like, I’d find it valuable if the interface encourages people to keep updating their probabilities later as they change. E.g., some (preferably optional) way of tracking how your view has changed over time. Probably also make it easy for people to re-vote without checking (and getting anchored by) their old probability assignment, for people who want that.
Note that Paul Christiano warns against encouraging sluggish updating by massively publicising people’s updates and judging them on it. Not sure what implementation details this suggests yet, but I do want to think about it.
https://sideways-view.com/2018/07/12/epistemic-incentives-and-sluggish-updating/
Yeah, strong upvote to this point. Having an Arbital-style system where people’s probabilities aren’t prominently timestamped might be the worst of both worlds, though, since it discourages updating and makes it look like most people never do it.
I have an intuition that something socially good might be achieved by seeing high-status rationalists treat ass numbers as ass numbers, brazenly assign wildly different probabilities to the same proposition week-by-week, etc., especially if this is a casual and incidental thing rather than being the focus of any blog posts or comments. This might work better, though, if the earlier probabilities vanish by default and only show up again if the user decides to highlight them.
(Also, if a user repeatedly abuses this feature to look a lot more accurate than they really were, this warrants mod intervention IMO.)