As of market creation, it's at 2.
@Evansbot is the bot in question.
Anything that halts trades when the bug is detected, reorders trades, or otherwise gets special handling counts. I'll post updates occasionally but won't trade. Follow the discord for the latest news.
Fixed bugs don't count, even if the code to detect / avoid is still active.
Update 2025-11-06 (PST) (AI summary of creator comment): The creator will not trade on this market (after setting the initial probability).
Update 2025-11-07 (PST) (AI summary of creator comment): Additional examples of what counts or doesn't count:
Yes+No shares in the same market/outcome would count if encountered and special handling code is written
Share estimate disagreements on multiple choice markets: Would count only if determined to be a Manifold bug (not a bot bug) and a workaround is implemented
Dependent multiple choice market positions (yes-in-all or no-in-multiple): Counted as a missing feature, not a bug
API returning incorrect data about share counts/balances: Would count as a bug if data is provably incorrect (not merely stale), but unlikely to get a workaround soon. Sufficiently stale data would count, but that's a high bar
1,000@Eliza Time period is up to you and how you want to do math; take a look at its trading history.
In addition to the two limit order related workarounds counted in this market so far:
There's an active bug with no workaround where in some circumstances not involving limit orders I get very slight disagreement between my share estimates and Manifold's delivered values on multiple choice markets. I'm not sure what the cause is, or whether that's a Manifold bug or one of mine. No workaround yet, would count if I decide it's a Manifold problem and implement a workaround (would involve discussion on the discord before I decided to count it).
I have code to deal with the fact that it's possible to hold yes-in-all or no-in-multiple on dependent multiple choice markets, but I'm counting that as a missing feature rather than a bug. Yes+No in the same market/outcome I'd count if I encounter it and write special handling code.
Sometimes the API returns incorrect data about share counts and account balances; I think I have one instance (didn't take enough notes) where that was not merely stale but impossible based on any reordering of recent API calls including the one that checked the account state. That seems transient and low impact, so it's unlikely to get a workaround any time soon. I'd count it as a bug if the data isn't merely stale; not sure where I stand on stale data, given no published guarantees. Sufficiently stale data would absolutely count, but that's probably a high bar.