Since this comment gets some attention every now and then, I'm leaving another comment here to point to this sequence where I flesh out this kind of view more.

Yes, or we don't need to have any specific reason to believe they do better than random. I think this could be consistent with phenomenal conservatism.

The second one is more about the grounds for justification (having warranted beliefs). Maybe judgement calls don't need to tend to be correct or for there to be the right kind of fit towards calibration for the resulting beliefs to be warranted. Maybe just the fact that something seems a certain way, e.g. even direct intuition about highly speculative things like the far future effects of interventions, can justify belief.

EDIT: This could be consistent with phenomenal conservatism.

FWIW, I don't think P1 and P2 together logically imply the conclusion as stated. I think you're probably leaving out some unstated premises (that might be uncontroversial, but should be checked).

For example, could anything other than evolutionary pressures (direct or indirect) work "against individuals unable to make the correct judgment calls regarding what actions do more good than harm (in expectation) considering how these impact the far future"?

Now you might say no, because our judgement calls are outputs of systems built up through evolution.

But I think an additional premise should capture that. It is not a tautology, but (possibly) an empirical fact.

Another: does making correct judgement calls enough to have warranted beliefs (in humans) about something require any (past) pressure against incorrect judgement calls (about those things in particular, or in domains from which there is enough generalization to the particular things)?

I think you'd say yes, but this is also an empirical claim, not a tautology.

Hmm, I might be misunderstanding you, so let me know if this doesn’t respond to what you have in mind.

I don’t think I've assumed any intervention is definitely positive in the long run. I have assumed they are each net positive in expectation for at least one set of targets (or types of effects) we can identify ahead of time (just humans, just farmed animals or just wild animal).

I guess I didn't specify how far these sets of targets go into the future, so you could assume I'm ignoring far future moral patients, or we can extend the example to include them, with interventions positive in expectation for some sets of far future targets.

The example is just illustrative, so I didn't want to complicate it with more than 3 sets of targets.

Coming back to this, on what timeline do you expect this kind of growth in wealth to happen, making animal welfare extremely cheap?

Or, with what probability would it not happen for at least another 15 years (and us not all dying for at least that long)? I'd guess 15 years is long enough for many animal welfare interventions to have significant impact, although on the shorter end. Some take several years to have any welfare impacts at all.

I'm imagining we could just discount welfare impacts by such probabilities. Animal welfare could still look quite cost-effective even after that, but it'll depend on the probabilities.

My first impression is that these techniques are pretty obscure and technical, and charities would not think to use them or know how to by default. In fact, sharing them here might make it more likely that charities use them (an infohazard).

EDIT: But maybe if motivated and strategic enough, they would find them through online search.

Were they successful in getting evidence removed from web archives?

Didn't they already address this specific vulnerability with the measures described on that page?

However, we remain concerned that in the case of a dispute, we would be accused of creating fake screen recordings/archives. 

If there is a third-party service that is trusted by the community that could verify the accuracy of our screen recordings/archives prior to us showing reviews to charities, we’d be much more open to the idea of showing reviews to charities before releasing them. Please let us know if you’re aware of one. 

https://web.archive.org/ seems good enough to me in most cases?

I think it doesn't do so well for Google Spreadsheets or videos, though:

1. For example, this archive of ACE's CEA of ALI only includes the first tab of the sheet. I tried seeing if I could archive the other tabs via their separate links, but it just redirects me here, where only the one tab is archived. Maybe there's a way around this, or a better acrhiving service.
2. AFAIK, videos are not archived at all, either.

I've also used https://archive.ph/. There's a browser plugin Archive Page for it, but I often get nginx errors when I try to archive pages with it. You can archive specific tabs by the tab links, e.g. here, but then it doesn't let you scroll through the sheet, which means you won't be able to access sheet cells you'd have to scroll to see. It also doesn't show how cells are calculated.

You could just save the whole sheet and upload it somewhere with a timestamp, and save archives anyway. Maybe there are better options than web.archive.org and archive.ph.