Hacker News Reader

Comment by rideontime | Hacker News Reader

Will Erowid and/or its users be at all compensated for providing the data for this?

"All information on Erowid.org is provided non-commercially for purposes of teaching, scholarship, research, and criticism."

The point of the website is to collect information about drugs for education and research purposes. It's a 501c3 nonprofit, which operates explicitly for public education. It's funded by donors, not by consumers of the info collected.

Hm odd, none of these seem to include ai “training”. It allows for human teaching but that’s far from ingesting data in a token database.

Research is very clearly covered under fair use.

Research falls under "purpose and character of the use", which is only one of the four factors considered in fair use analysis. I gather a court could decide that this outweighs the other factors, but I doubt that can safely be taken for granted as a general rule.

Even when you do it to develop a commercial product?

As someone who contributed to the site many years ago, I am more than happy for my experiences to be used in this fashion.

I second this.

Did you ever use the site? It was like a handshake agreement situation where you had no idea how full of shit the posts were. If people weren't outright lying, they could (very well) be experiencing wish fulfillment or placebo. Many of the "negative" experiences were similar - describing reactions that were frankly hard to believe. Posts tended towards fictional exaggeration. Further, many users were anonymous, pseudonymous, lying about identity, or posting in a time when many believed no one would be probing through the internet trying to identify folks.

tl;dr - I seriously doubt any of the users wants to be identified, much less compensated for providing "data".

The trip reports for datura are legendary for their outlandishness.

On point though, many of the chemicals in question aren't legal so not sure how much users would want to be identified..

If you had actually ever tried datura, there's no way you'd be so dismissive of people's reports of using it as being "outlandish".

Yes, no true scotsman would dismiss a datura trip report.

strong agree

Datura is absolutely outlandish, though.

I'm a very experienced psychonaut and even I won't dare touch that stuff. One of my friends did and just watching him on it made me never, EVER want to do it.

That's been my experience with the site, too, which made me raise my eyebrow at the article's claim of "first-hand testimonials that have been vetted for authenticity and accuracy." If the data's such junk (which I surely won't argue with), why study it?

e: From the original paper:

> Rigorous quality control is conducted by trained experts with careful consideration given to “quality, credibility, and focus on effects or outcomes” (www.erowid.org). By means of that mechanism, all reports that provided the basis for the present investigation have successfully exceeded a standard quality filter.

Totally.

I once had a disagreement with a lead dev for an in-house tool that was a glorified DAM. their forms were not kept up to date, and so users were forced to enter bad data which skewed query results. when pressed about why the fields were forced to choose from an outdated pre-populated enum, i was told "bad data is better than no data".

maybe the AI researchers are from the same school of thought. having bad/unvetted medical data/advice on psychedelics would be better than no data at all, right? /s

They will get a reputation bonus.

Thankfully, outside of Hacker News, and particularly on Erowid, there is not as high a frequency of the “where’s MY payday” mindset. This is just American / Silicon Valley brain rot.

But then why do we have to pay for accessing ai content? Can we download the models openai and others made using people’s work?

Reply on news.ycombinator.com