Still Andrew (I guess) is a user on You can follow them or interact with them if you have an account anywhere in the fediverse.
Still Andrew (I guess) @ajroach42

I'm not doing another re-captcha.

If I'm trying to do something, and you require me to do unpaid labor for google, I'm just going to 1) not do the thing 2) publicly ridicule you.

· Web · 9 · 13

@ajroach42 Also, how many fucking cars/storefront/buses do I have to recognize before I can do what I came here to do .... This bs is really getting out of hand :/

You can bypass all that RECAPCHA nonsense by going to the site using the Tor Browser. I don't know how they do it, but every site that requests a RECAPCHA in other browsers doesn't in the Tor Browser.

@ND3JR I see still see recaptchas in tor browser often enough for it to be annoying.

Unless something has changes in the last two weeks, I think you must be mistaken.

@ajroach42 No mistake, at least when it comes to browsing Freethought Blogs using Tor Browser. I do have a separate Tor daemon running and I also go through Privoxy, which goes through Tor.. Maybe it's that combination that works for me.

@ajroach42 Hear hear, and let me just name & shame Patreon here -- a terrible company whose business model consists of creaming off ever more money from people of good will trying to help each other -- but can't be bothered to do proper account security. They just farm it out to Google's recaptcha, thus inconveniencing the PEOPLE WHO SEND IN MONEY.

@ajroach42 maybe a stupid question, but how exactly does recaptcha translate to unpaid labor ?

@pranav Google uses the new visual re-captchas to train machine learning algorithms.

We're playing "spot the fire hydrant" and "spot the traffic light" and "spot the store front" so that google can feed that data in to a big ol' database, and use it for profit.

This has more or less always been the case with re-captcha, although originally it was supposed ot be helping OCR books (I'm all for OCRing books! I might even do one or two words a month for free and not mind. This is different.)

@ajroach42 @pranav I feel like "spot the fire hydrant" could still be used to make better image captions automatically for blind people

@teascade @pranav Sure, it probably could.

But it can also be used for googles Self Driving Car Program, or their drone program, or whatever else they decide to do with our free labor.

And I won't assume best intentions of google, because we both know they don't have best intentions.

Are there any viable alternatives?
I've heard that bots are getting up to 50% success with re:captcha now.

@ram I saw a workaround for re-captcha specifically with an 85%+ chance of bypass, using google's own tools.

The viability of workarounds depends on your likely threat model, and what it is you're placing the barrier in front of.

Dang. I guess this will always be an arms race.

@ram That's the point though.

re-captcha is like DRM. It doesn't actually prevent the thing it's supposed to prevent, and it has all kinds of unintended side effects.

And that's without getting in to recaptcha v3, which is all tracking/spying based.

We need more thoughtful ways to deal with bad actors.

Ultimately, I think the only way to prove that a person is a human is to have a human administer the admission process. But even that is problematic, because the Turing test is gradually crumbling.

@ram I think the "is this person a human" model is broken.

There are some valid reasons for bots to access services, and some invalid reasons for people to access them.

Heck, humans can be more abusive to a system than machines, when properly motivated.

Plus, services like Mechanical Turk let you outsource captcha bypassing for fractions of a cent per captcha.

They've been infective for years.

If the goal is stopping bad behavior, then captchas are a long outdated solution that should be replaced.

With what? It depends on what the captcha is in front of, what bad behavior it was supposed to be preventing.

There isn't a single solution, there never was, we've just been pretending.