Some things I noticed about google re-captcha already. It can't tell the difference between a 3x3 matrix of a house and a bus, it doesn't know the difference between scooters and motorcycles, it has a margin of error on fire hydrants/usually the last image that it asks you to identify a fire hydrant in and fades out will not bring up another fire hydrant so you can click verify early.
Human-Assisted Archival of Yahoo Groups
11–20 of 65 posts
Re: Human-Assisted Archival of Yahoo Groups
#12Jumped in when we were joining groups starting with 'S', and now we're already at 'E'. Really satisfying. Given that my account got a connection attempt from Sweden, I guess that's where the archivist live. Hopefully he will have a nice morning tomorrow thanks to the community.
Are you going in reverse alphabetical order, or what?
Re: Human-Assisted Archival of Yahoo Groups
#13Jumped in when we were joining groups starting with 'S', and now we're already at 'E'. Really satisfying. Given that my account got a connection attempt from Sweden, I guess that's where the archivist live. Hopefully he will have a nice morning tomorrow thanks to the community.
Are you going in reverse alphabetical order, or what?
I have read on the other hn thread that this is the case
Re: Human-Assisted Archival of Yahoo Groups
#14Wonder if we could make a deal with all those casino seo spammers from indonesia. (Or whoever sells their service to them.) If they will use their cache of yahoo addys to exfiltrate Yahoo Group content, we'll give them a free month of gitlab user spam usage no questions asked.
I agree there must be something that could be done to utilize the click farms elsewhere.
Re: Human-Assisted Archival of Yahoo Groups
#15Could Mechanical Turk be used to do this? I'd be happy to make a donation to help offset the cost.
Re: Human-Assisted Archival of Yahoo Groups
#16Some things I noticed about google re-captcha already. It can't tell the difference between a 3x3 matrix of a house and a bus, it doesn't know the difference between scooters and motorcycles, it has a margin of error on fire hydrants/usually the last image that it asks you to identify a fire hydrant in and fades out will not bring up another fire hydrant so you can click verify early.
Recaptcha doesn't know the difference between anything. It's just trying to gather consistently tagged training data of regions which can in turn be used to train something that can identify objects.
Crowdsourced labeling doesn't make sense if you already have something that can tell you what the label should be.
Re: Human-Assisted Archival of Yahoo Groups
#17I wonder if groups with actual members will be OK. I'm a member of one and have several years of message digests in my email. I just downloaded all of them. For groups without members, or with no-one bothering to read them anyway, maybe it doesn't actually matter?
Re: Human-Assisted Archival of Yahoo Groups
#18Can you only join one group per account?
Re: Human-Assisted Archival of Yahoo Groups
#19Is there a possibility of asking ReCaptcha to disable their protection for Yahoo Groups? It's a reach I guess.
That's Google, and that would be suicide to the brand.
Re: Human-Assisted Archival of Yahoo Groups
#20Is there a possibility of asking ReCaptcha to disable their protection for Yahoo Groups? It's a reach I guess.
Thinking like a different kind of hacker. I like it.