It seems to me like the existence of @acc negatively impacts user experience, especially by punishing errors, while not providing any significant value.
Additionally it seems to have access to at least calibration, which is not part of API, and this has an unfair advantage over other bots (and in practice humans).
The only upside to its' existence I see is extracting Mana from circulation, but the amount does not seem very significant in the total scale.
I can be reached on discord if you want to discuss an argument in advance to limit your risk / maximize profits.
I wonder if having stocks about different things on Manifold would be useful to help the devs know which features work the best/are most well-liked. I made this market to find out what most traders think of @acc
@JosephNoonan not a "stonk" like that. If any company actually believed stocks like that were meaningful, I'd short the company's stock so fast lol
The better way to do this is a poll. Manifold does run many of those and they provide useful feedback
Maybe a better way to think about acc is just as an extension of the AMM. By default, there is a fixed equation for how much liquidity you can get for a given trade determined by the AMM parameters. But sometimes, Manifold's house account decides to offer you a little extra liquidity on top of that (or a little bit less liquidity, for future traders) based on their opinion of what a fair price for the market should be after your trade. Other than being a bit less transparent than the regular AMM (kind of unavoidable to prevent it being exploited) it seems like a pretty reasonable system that empirically does improve prediction quality.
If Manifold is a place for getting accurate predictions, and @acc is making a consistent profit across many markets, it means that @acc is contributing to more accurate predictions. I think that's good for Manifold.
I think very few of @acc's trades are on big-time "oops" user mistakes. I'm not sure how much of @acc's profit comes from such mistakes.
If Manifold had better controls for undoing a trade that a user would immediately realize was a mistake, would you see a lot less downside in @acc?
@Charlie "[if] @acc is making a consistent profit across many markets, it means that @acc is contributing to more accurate predictions", acc making a profit doesn't seem to necessarily imply that it's making the markets more accurate, it's just siphoning money away from people on either side of a question every time they bet, and making the markets more "accurate" insofar as it's marginally reducing noise by punishing trading
but the counterfactual i'm comparing it to in my mind is a different trading mechanism that does not require @acc to exist to reduce noise. for instance, a mechanism that drips mana onto a side at a certain rate, or more generally has parameterized response curves to market events created to minimize noise without siphoning mana away from users potentially with every single action (which may as well just be taxes in some sense, if manifold wants to go that route)
@Charlie a lot less downside, it would still siphon away profit margins from real traders, making it harder to make a profit and thus in my opinion most likely negatively impacting user experience -> retention -> growth
I do agree that it most likely makes markets more accurate, or at least less noisy, but I don't think that matters now in comparison to said user experience which it (and in general bots of that kind) make worse
@MartinRandall I'm not 100% sure TBH, probably overstated "a sufficient delay on API and taxes could easily make high frequency trading type activities impractical"
@CodeandSolder As Twitter has allegedly found, if you charge/delay too much for API use, then you get screen-scraping.
@MartinRandall this is in my opinion a much larger issue for platforms like Twitter than here because of scale and no initial investment required. MM could probably detect it and ban the account often enough it would not be net profitable.
the mana collected by @acc should be used to pay the league prizes. Unless league is being funded by something else...
@higherLEVELING What does it even mean for the league to be "funded". Manifold can print mana at will! It doesn't cost anything!
@VitorBosshard i understand, but i just wanted to give ACC an actual reason for being there. If they can just keep printing mana, then why have the bot come around and steal ours? we're getting taxed without seeing any benefit
I think part of the impetus behind creating @acc was that SG saw @Botlab and @v competing on speed, and decided that it would be better for Manifold if bots competed with each other on price accuracy rather than speed alpha, so he put an end to the race by building the definitive speed bot. Acc can place trades in between two halves of a double-sided limit order.
But now we have six different arbitrage bots racing each other. We need arbceleration.
@PatMyron combining duplicate markets with current incentive structures seems difficult, as there's lots of competing incentives that don't seem to be able to be merged. (trader bonuses, different pricing mechanisms for market types)