resolved Mar 20
Was it an unpleasant surprise when trades made by market creators became retroactively public?
Resolved
N/AThis market is a pure opinion poll and will resolve N/A. Comments encouraged.
💬 Proven correct
@Austin: If you're worried that stuff like the leaderboards expose information about individual trades even if all those trade are private, you can make the leaderboards fuzzier in a few ways without hurting their core functionality:
- rounding numbers to a few significant digits
- updating them at a set time (e.g. daily), rather than in real time
(and other options I'm not thinking of)
To give one example, the game client Steam at one point accidentally leaked precise player numbers for all games sold on its platform because its "how many players have completed this achievement" stat yielded 16 digits of precision if queried via its API: https://medium.com/@tglaiel/using-achievement-stats-to-estimate-sales-on-steam-d18b4b635d23
Steam solved this problem by having its API output rounded numbers instead.
💸 Best bet
A trader bought Ṁ800 YES from 12% to 67%
Gurkenglas made Ṁ354!
Sort by:
(TIL that there's a way to make a small profit if you believe a market will definitely resolve N/A.)
By the way: This change has now been undone; trades by the market creator should be as anonymous as trades by other people.
@Austin: If you're worried that stuff like the leaderboards expose information about individual trades even if all those trade are private, you can make the leaderboards fuzzier in a few ways without hurting their core functionality:
- rounding numbers to a few significant digits
- updating them at a set time (e.g. daily), rather than in real time
(and other options I'm not thinking of)
To give one example, the game client Steam at one point accidentally leaked precise player numbers for all games sold on its platform because its "how many players have completed this achievement" stat yielded 16 digits of precision if queried via its API: https://medium.com/@tglaiel/using-achievement-stats-to-estimate-sales-on-steam-d18b4b635d23
Steam solved this problem by having its API output rounded numbers instead.
Austin: other way around! Restricted group markets should be private bets. Your social group is probably already selected for nondefectors such that you don't need to audit them much. The ingroup's *opinion* of what you bet, otoh, is a strong distortionary factor to personal markets.
re: Austin: I think both are separately useful. I've wanted private markets for the purpose of not making the *existence of the market* public (but it still seems fine to me to have trades in those be anonymous). but also even in public markets that aren't dealing with anything too confidential to publicly name it's useful to have anonymous trading - I would be so much more hesitant about trading even in current events markets if I felt like every time I made a trade I would have to be able to justify that trade publicly.
Thanks for the feedback! I want to disentangle between two kinds of privacy:
1. "Anonymous participation in a public-facing market"
2. "Market restricted to your team or community or circle of friends, but identities are visible within"
So far I'd mostly been considering the first kind of privacy, but it seems like 2. would be a better fit for @BlazingDarkness's use case? That would be good to know, since it changes what we build.
(Of course, these aren't necessarily mutually exclusive; I could imagine wanting both. But on a technical & product side they lead to different designs.)
My opinion is that if you guys *do* start to include more settings, then a private/public toggle per market would be a good choice for one. There seem to be enough pros and cons that both types of market would be valuable, and I personally would get more use out of it than, eg, free response markets, which as currently implemented I just will probably never create or bet on; I don't really grok them 😔. Possibly that sort of thing could be in an 'advanced options' expandybox, with one or the other being the site-wide default, to lessen the number of decisions needed to create a new market? but of course IANA dev and this might be in some way horrible; this is just from my experience as a user.
+1. At least with leaderboards we mostly still get plausible deniability.
strong agree with Elena - there's a lot of markets I'd feel much less comfortable betting in if it had to be public. this info being theoretically accessible through leaderboards is much less bad.
Thanks for the insight, Elena! We do want to make users feel comfortable placing bets on personal and controversial topics here on Manifold. Anonymity does trades off against some features that we'd like to have, though; e.g. even today you can glean some information about who placed which trade just by looking at the total and per-community leaderboards. (But there is a big difference in user experience between having aggregated public info and specific revealing info...)
Possibly one thing way to thread the needle is to explicitly have any particular bet be anonymized (and if it is, we exclude it from leaderboard tallies as well). The downside is more "settings inflation" or UI complexity, but perhaps we can manage this with a general user setting.
I think trading being and staying anonymous is important for people's ability to make honest trades about lot of topics! If identities are attached, then people have to weigh the costs of revealing their potentially unpopular thoughts against possible M$ gains, and I expect for personal and controversial topics that will result in fewer trades.
That's a pretty good summary of the costs and benefits IMO. If all trades eventually become public it will be really awkward to buy NO on someone's "Will I stay with my current partner" or "Will I get a good performance review".
One more thing: I'm currently considering making all trades public once a market has resolved. The pros of this would be:
- Allow a per-market leaderboard
- Insight into where traders have been making good decisions
- Auditability against cases of fraud or manipulation
Cons:
- May unpleasantly surprise traders
- May have a chilling effect on unpopular bets
This could be done retroactively for all markets; or only on new markets going forward; or only after a period of delay, so traders have more time to benefit from insider information. Would love to hear your thoughts!
Hey! I'm the one who implemented this change, and I'm sorry that the rollout came as an unpleasant surprise. On reflection I do agree that this was very sudden; if I were to do this over, I'd probably only display creator trades moving forward, perhaps after an announcement period. (At the time when I shipped, I hadn't paused to consider the privacy costs.)
It can be difficult to strike a balance between "iterating quickly" and "upholding implicit commitments"; here I think I messed up. Thanks to everyone for pointing this out; we'll aim to do better in the future!
Don't think there can be said much more, it seems like almost everyone agrees that the retroactive part is the problem
I don't like that it retroactively affected existing trades. But it's probably good for future trades.
I assumed that this was public already? And it was, via API.
Being able to assess market manipulation by the creators is good going forward. But yeah it was very unpleasant to suddenly have to consider the optics of old uncommented bets.
Making anything retroactively public is unprincipled and sets a bad precedent.
Wanted to drop the percentage a bit because I think a lot of people weren't bothered. That said, I do share the intuition that what was posted as private should stay private.
I didn't particularly have anything to hide, and I know people could already see things through the API, but this was still kind of a nasty surprise for me! As a bit of a lurker by nature, I strongly prefer to know in advance which things I do will be public or private. Obviously it would have been more work for the devs (appreciate you so much, devs <3), but I think this should have been implemented differently, possibly in the manner I suggested on the discord a few hours ago:
"I maybe would keep existing trades as they were, put a little 'ℹThis market was created before creator trades were public' piece of text somewhere when it applied, and similarly put something like 'ℹ Trades are public for market creators' somewhere in the box when you're buying a bet on your own market."
A trader sold Ṁ11 NO from 83% to 84%
A trader sold Ṁ1,173 YES from 91% to 83%
A trader bought Ṁ2,000 YES from 67% to 91%
A trader bought Ṁ800 YES from 12% to 67%
A trader sold Ṁ692 NO from 5% to 12%
A trader bought Ṁ1,000 NO from 21% to 5%
A trader bought Ṁ500 NO from 92% to 21%
A trader sold Ṁ50 NO from 75% to 92%
A trader bought Ṁ50 NO from 92% to 75%
A trader sold Ṁ1 YES at 92%
A trader sold Ṁ7 YES at 92%
A trader sold Ṁ22 YES from 93% to 92%
A trader sold Ṁ23 YES at 93%
A trader bought Ṁ300 YES from 62% to 93%
A trader bought Ṁ1 NO from 63% to 62%
A trader bought Ṁ10 NO from 69% to 63%
A trader sold Ṁ0 YES from 70% to 69%
A trader bought Ṁ20 YES from 63% to 70%
A trader bought Ṁ1 YES from 62% to 63%
A trader bought Ṁ1 YES at 62%
A trader bought Ṁ32 NO from 91% to 62%
A trader bought Ṁ1 YES at 91%
A trader sold Ṁ11 YES from 92% to 91%
A trader bought Ṁ10 YES from 91% to 92%
A trader sold Ṁ5 YES at 91%
A trader sold Ṁ1 YES from 92% to 91%
A trader bought Ṁ5 YES from 91% to 92%
A trader bought Ṁ1 YES at 91%
A trader bought Ṁ1 YES at 91%
A trader bought Ṁ1 YES at 91%
A trader bought Ṁ1 YES at 91%
A trader sold Ṁ1 YES at 91%
A trader sold Ṁ1 YES at 91%
A trader bought Ṁ1 YES at 91%
A trader bought Ṁ1 YES at 91%
A trader bought Ṁ40 YES from 82% to 91%
A trader bought Ṁ1 YES at 82%
A trader bought Ṁ1 YES at 82%
A trader bought Ṁ5 YES from 80% to 82%
A trader bought Ṁ20 YES from 69% to 80%
A trader bought Ṁ3 YES from 67% to 69%
A trader sold Ṁ22 YES from 80% to 67%
A trader sold Ṁ1 YES at 80%
A trader bought Ṁ1 YES at 80%
A trader bought Ṁ20 YES from 69% to 80%
A trader sold Ṁ17 NO from 46% to 69%
A trader bought Ṁ5 YES from 40% to 46%
A trader bought Ṁ20 NO from 65% to 40%
A trader bought Ṁ19 YES from 37% to 65%
A trader bought Ṁ1 YES from 34% to 37%
YES shares
2198
43
23
