From techmeet
back
Third hour
Sep 30 17:00:18 boud ouch :(
Sep 30 17:00:56 dannyp they are not the only solution for dealing with spam, nor are they a permanent solution - there are already reports of companies in economically oppressed portions of the global south using human labor to solve captchas
Sep 30 17:01:07 kwadronaut dannyp: i forgot, which imc are you from? I just doubt that it *is* adapted to visually impaired people, so i would like to test or try that out.
Sep 30 17:01:13 ryan dannyp: oh come on
Sep 30 17:01:22 ryan thats the only argument against captcha's
Sep 30 17:01:40 ryan i believe in blind publishers
Sep 30 17:01:48 ryan i believe they can find a friend or something to help them with the captcha
Sep 30 17:01:55 ryan blind indymedia publishers are a resourceful group!!!!
Sep 30 17:01:57 ryan they can problem solve
Sep 30 17:02:20 dannyp kwadronaut: pittsburgh, and it could be better, you're right. but it's functional in lynx
Sep 30 17:10:11 dannyp it's just that there are solutions that are better... solutions that deal with right-wing trolls as well as bots... if linkspam is all you want to defeat, just strip URLs out of untrusted comments. Or replace them with a tinyURL-like alias on the IMC until they can be approved/rejected
Sep 30 17:11:10 dannyp perhaps only put URLs that aren't to somewhere inside the indymedia.org domain through that treatment
Sep 30 17:12:13 kwadronaut dannyp: that last suggestion isnt a good one, trolls like to point to their hidden things andso
Sep 30 17:12:31 dannyp kwadronaut: hmm?
Sep 30 17:12:37 dannyp oh, yes
Sep 30 17:12:47 dannyp the URL comment was for dealing with linkspam, not trolls
Sep 30 17:13:13 dannyp trolls are best dealt with with 1. easier hiding of comments (inline administration), and possibly 2. bayesian analysis
Sep 30 17:13:21 dannyp CAPTCHAs don't even begin to address trolls
Sep 30 17:13:28 kwadronaut it's not any easy struggle, prolly need to compromise anyway
Sep 30 17:13:41 kwadronaut that is the struggle against unwanted content in geral
Sep 30 17:13:57 kwadronaut (trolls + linkspammers)
Sep 30 17:14:08 kwadronaut some trolls are spammers though ;-)
Sep 30 17:14:35 dannyp but I feel there are technical options - many of which haven't been explored well - that help deal with both trolls+linkspam, and that CAPTCHAs are an easy way out of half of the problem, that create a whole new set of difficulties
Sep 30 17:14:51 dannyp in that it's a step back for accessibility
Sep 30 17:15:15 dannyp and captchas aren't even a long-term solution for linkspam
Sep 30 17:22:23 txopi thay are captchas not based on images that are very good
Sep 30 17:22:32 txopi text /human based ones
Sep 30 17:22:44 txopi and they are good in terms of accessibility
Sep 30 17:23:47 txopi the meeting is end?
Sep 30 17:23:58 dannyp yes, the meeting ended awhile ago. we're just talking now
Sep 30 17:24:04 txopi ok
Sep 30 17:24:18 dannyp text captchas are better. but they too will be easily overcome in time, especially by the anti-captcha industry
Sep 30 17:24:35 txopi i don't think so
Sep 30 17:25:08 dannyp the reality is, for as long as corporations can pay people a few cents an hour to solve pages full of captchas at a time, that captchas won't be an acceptable solution
Sep 30 17:25:21 txopi "wrety ni txet vife pals vife" -> ten
Sep 30 17:25:37 txopi "wrety ni txet vife puls vife" -> ten
Sep 30 17:25:38 dannyp and I'm all for overthrowing global capitalism such that nobody feels like selling their labor for so little is worthwhile, but we're not there yet
Sep 30 17:26:55 dannyp god, it took me about 2 minutes to understand what that meant, txopi
Sep 30 17:27:04 Zapata but... dannyp
Sep 30 17:27:04 Zapata if they have to pay people a few cents an hour
Sep 30 17:27:04 Zapata that constitutes a barrier...
Sep 30 17:27:13 txopi :-D
Sep 30 17:27:53 dannyp yes, that's true. but it's not foolproof, and the industry is growing, and I don't think we're far from seeing more and more linkspammers willing to pay that
Sep 30 17:28:53 txopi "Does the human mind raed wrods as. a wlohe?"
Sep 30 17:30:02 txopi the system used by akismet and spam-karma (in wordpress) is very efective too
Sep 30 17:31:39 txopi i agree there is a war between spammers and people that want to avoid it, but i think corporations don't have a easy way at all
Sep 30 17:32:02 * dannyp notes that soundex('write') = soundex('wrety'), soundex('plus') = soundex('puls'), and the other words are simple anagrams
Sep 30 17:32:03 txopi well, is just my opinion
Sep 30 17:33:05 dannyp anyhow, my point is, don't allow CAPTCHAs to make you lazy about implementing real, long-term solutions for content regulation
Sep 30 17:33:13 dannyp and to remember that CAPTCHAs are not accessible
Sep 30 17:33:21 dannyp (many of them aren't, that is)
Sep 30 17:33:57 txopi wich word doesn't fit here? t*ble, door, planet, house, wind*w
Sep 30 17:34:18 jebba planet! :)
Sep 30 17:34:21 txopi :-)
Sep 30 17:34:27 jebba what do i win?
Sep 30 17:34:48 txopi you are the one who is going to code all the new cms for indymedia
Sep 30 17:34:50 Zapata you may now submit a posting to indymedia for free!!!!!
Sep 30 17:34:52 txopi woooow!
Sep 30 17:34:53 txopi :DDD
Sep 30 17:35:00 jebba maybe we could have a spam defeating obstacle course and see who is fastest ;)
Sep 30 17:35:14 dannyp I'm not even certain why planet doesn't fit
Sep 30 17:35:25 jebba solve captchas, audio captchas, turing tests, etc....
Sep 30 17:35:31 kwadronaut i still find it a great idea to have a *publish on all indymeidias* ;-)
Sep 30 17:35:41 kwadronaut *button
Sep 30 17:35:56 dannyp hah
Sep 30 17:36:21 dannyp well, making better use of syndication will hopefully reduce the desire of non-spammers to post to multiple IMCs...
Sep 30 17:37:02 gdm we live in a world of hope :)
Sep 30 17:37:40 dannyp ...or to put it another way, if we make better use of syndication, hopefully collectives will feel comfortable hiding things that don't belong on their local
Sep 30 17:39:00 ryan dannpy: if anything, user accounts are an increased measure against link spam
Sep 30 17:39:06 ryan dannyp: because it allows personal filtering
Sep 30 17:39:09 ryan dannyp: and trusted users
Sep 30 17:39:18 ryan dannyp: which are more sophisticated solutions to spam than captcha's are
Sep 30 17:39:33 ryan dannyp: in the meantime, i'm in favor of whatever works
Sep 30 17:39:41 ryan dannyp: and captcha's are working for us (sf-imc)
Sep 30 17:41:18 txopi do you know how akismet and spam-karma works? the idea is great and they are really effective (so you don't event need captchas)
Sep 30 17:41:29 dannyp nope, googling them now
Sep 30 17:42:02 ryan txopi: this is what i'm talking about
Sep 30 17:42:11 ryan txopi: spam protection through identity keys (via user accounts)
Sep 30 17:42:19 ryan txopi: but its interesting to see these solutions you showed us
Sep 30 17:42:27 txopi aps
Sep 30 17:42:28 txopi sorry
Sep 30 17:42:36 ryan nah nah
Sep 30 17:42:39 ryan i never heard of these before
Sep 30 17:42:40 ryan i wanna see
Sep 30 17:42:45 dannyp what's nice with slashdot's model, is that anonymous cowards can still get modded up to be seen
Sep 30 17:42:56 dannyp were it not for modding up ACs, I'd never read an AC's post...
Sep 30 17:43:22 dannyp my concern is that any identity-based solution might make it harder for anonymous contributions to be seen
Sep 30 17:43:52 ryan *shrug* lets see the implementation first
Sep 30 17:44:26 dannyp no, let's design it properly first, so those doing the implementation aren't wasting their time
Sep 30 17:44:57 jebba i think we're more in hunter gatherer mode than building mode
Sep 30 17:45:12 * ryan agrees w/jebba
Sep 30 17:46:42 dannyp fair enough. but we need to know what's worth picking up and what we should leave behind...
Sep 30 17:47:58 dannyp our constraints against IP logging make some things harder for us. It won't surprise me if this is one of the few things we'll need to roll our own for
Sep 30 17:48:18 dannyp but it's not like we'd be the first to use bayesian analysis
Sep 30 18:08:40 ryan eh, spammers use tor
Sep 30 18:12:21 txopi i think that the most effective systems to detect spam (into email, blog's comments...) analize the content
Sep 30 18:12:31 txopi not agents, ip, etc.
Sep 30 18:13:32 ryan txopi: yes i agree
Sep 30 18:18:36 dannyp as do I. bayesian analysis!
