search

quantified-uncertainty / metaforecast

Fetch forecasts from prediction markets/forecasting platforms to make them searchable. Integrate these forecasts into other services.
https://metaforecast.org/
MIT License
56 stars 5 forks source link

Documentation for grapqhl usage #65

Open NunoSempere opened 2 years ago

NunoSempere commented 2 years ago

I want there to be some documentation for how to use the graphql api, with some simple instructions. I'm planning on writing it myself. I also want to there to be a post on the EA forum/LessWrong announcing it. I'm up for writting it as well and then either coauthoring it with you @berekuk or mentioning that you did the work on this. Before that, though, I'd love to have history added to the graphl (https://github.com/quantified-uncertainty/metaforecast/issues/64)

We could also wait until we have added the charts functionality and have a longer update post.

NunoSempere commented 2 years ago

You also mentioned that it would be good to have a mailing list for metaforecast, so I'll go ahead and create one right now on substack unlesss you have any objections. Would be good to mention breakages such as https://github.com/quantified-uncertainty/metaforecast/issues/46

NunoSempere commented 2 years ago

Thoughts?

NunoSempere commented 2 years ago

So I'm very bullish on releasing alpha functionality very quickly. On the other hand, that has not worked as well on Squiggle. Overall, I'd probably want to advertise the graphql a bit, but then maybe also create a mailing list to which users should subscribe to be alerted of breakages. Does that sound crazy?

berekuk commented 2 years ago

maybe also create a mailing list to which users should subscribe to be alerted of breakages

I'm worried that it might not be worth it at this point, i.e. the costs of notifying the mailing list about every change would be higher than the benefit to our currently non-existent users.

Let's wait until we get any API consumers and see how strict their stability requirements are? I hope we'll know them and their use cases personally, so we can just ask.

I'm up for writting it as well and then either coauthoring it with you @berekuk or mentioning that you did the work on this.

Yeah, let's do that!

berekuk commented 2 years ago

I'm worried that it might not be worth it at this point, i.e. the costs of notifying the mailing list about every change would be higher than the benefit to our currently non-existent users.

I've thought about this some more and turns out my aversion is mostly in "setting up a mailing list is costly" + "group emails should be written properly and it's a lot of work".

I'd be much more in favor of a telegram channel or a slack channel on EA Forecasting slack (but that one is invite-only).