4 ms·
Why your captcha isn't working - human armies of spam posters for hire
- ck2 15y agoAlso a good read: http://www.technologyreview.com/blog/arxiv/27357/ http://www.technologyreview.com/blog/arxiv/27357/
- codemac 15y agoThe future of online communication is based around these "filter bubbles". The more aware may see that this is a bad thing, but most people see this is a great improvement. I'll add my friends on diaspora, google+, facebook, etc, and we'll all send each other stories that make our lifestyles appear worthwhile. If you're battling captcha, you aren't fighting the battles of the future. Not that it isn't a worthwhile problem, I just see facebook's auto grouping, g+ circles and diaspora aspects as things that all make conversation socially verified ahead of time.
- bravura 15y agoThis. Personalization is the right way to combat spam.
- praptak 15y agoIt's rather "human trust-based networking" than personalization. And it rocks since the dawn of humanity as a solution to problems ranging from censorship (good luck censoring gossip), restrictions on trade (good luck stopping friend-of-a-friend drug deals), shady companies (bad opinions travel fast), hiring and yes, the good old spam too. I believe we have not yet realized the full potential of human networking + computer networking.
- ericb 15y agoWhat about a captcha system that costs a penny?
- jasonkester 15y agoI've been noticing this type of activity on my user-content sites for a couple years now. It'll make it past any human-detection script you throw at it, and it seemingly doesn't care that 99% of it gets blocked by bayesian filters and would have no-follow links on no-index pages even if it made it through. The only effective solution to human-powered spam is to flip your user-generated content from "public by default" to "probably spam unless proven otherwise by a moderator". Watch your users posting things, approve the non-spam to be publicly visible, mark consistently good users as "trusted" so that you don't have to watch them anymore. It takes more human intervention, but really that's the only way to make a user-generated content site work these days. (Unless, of course, your userbase is an army of geeks who are willing to upvote good stuff and flag spam for you). More here: http://expatsoftware.com/articles/2010/03/care-and-feeding-of-happy-spammer.html http://expatsoftware.com/articles/2010/03/care-and-feeding-o...
- atirip 15y agoWhat about to ask for Chinese censors to kindly add your forum to the Great Firewall? Problem solved.
- Tuna-Fish 15y agoThat made me start wondering what kind of content should I post to maximize the odds of that. ... But it wouldn't actually work. This isn't exclusive to China -- it's also appearing in India. And this won't go away until wages equalize globally enough to make this kind of things unprofitable. And that won't happen very soon.
- ck2 15y agoThat's an interesting idea. Post about Tank Man for China. India is a bit more tricky http://en.wikipedia.org/wiki/Internet_censorship_in_India http://en.wikipedia.org/wiki/Internet_censorship_in_India But in the end it would be unfair to regular citizens of those countries who have about as much control over their governments as the USA citizens do if you aren't a lobbyist or the 1%
- buro9 15y agoIt's not just the Chinese. My spam logs suggest it's also the Russians and Indians in great number and then there's a large gap before reaching the USA and some European countries. Besides, do you believe you'll never have any valuable users from those places? IP blocking is proving to be the most successful thing I've got going. I do like to make it forget IPs after a while as they could just be temporary. Most of my IP blocks I apply for a month, it's an entirely arbitrary time period I know. I did do an experiment and captured all registrations and form submissions for a day and looked at the data. The only pattern that ALL spammers that day could be identified by was their password. The password was either '0' or '[a-zA-Z0-9]{10}'. Unfortunately I can't just block those patterns as I have no idea how many real users would be affected by such a thing. Obv' I don't have their passwords and just store hashes, and I don't normally transmit a password except for noscript graceful degradation. So basically... whilst I was pleased to find a pattern it's not one I can implement against without doing something undesirable elsewhere.
- paulhauggis 15y agoIf you've ever looked at some of the black hat seo forums, this is a method they use for "marketing".