So, they are back. Okay muchachos, pay attention. This is relevant. I've spoken with some of my contacts at certain Other Governmental Agencies. This is not the work of the Chinese. The spam originates from a Russian Troll Factory in Minsk. Vladimir Putin is angry that CB upset @Aldo, which then upset La Abuela de Aldo, a woman for whom Vladimir has "intentions".(as noted in the April Speculation thread, https://forum.corvusbelli.com/threads/april-come-she-will.1786/page-89#post-43755) Of course, Vladimir showing up at her front door, unannounced and unexpected, stinking of vodka and borscht, this probably upset her more. Or that he was not wearing a shirt at the time. Or pants. He's also kind of pissed that the USAriadna sectorial released before the Tartary sectorial. Thus, his oligarchs have unleashed the trolls. Gives them something to do until the next US election.
Do the timezone maths... and nothing better at half morning than some CB sweat to drink. *tongue-out* Ohh, WAIT! The Koreans are into this too now! *bwahahaha* Online gaming spam. Makes sense, no sectorial for them, they have to resort to casinos if they want to roll some dice.
Soooo... how long until the JSA spambots arrive to fight the YJ ones - by outspamming their spam with their own spam..?
There is a joke that goes like: "I revoked my private security contract, I donated all my firearms, and decorated all of my house's exterior with pro-yihadistic, nazi, and cartel-like symbols. So now I have the FBI, the NSA, the police, the DEA... providing me with security 24/7, for free. I've never felt so sure!" As for the "kindergarten for newbies", sadly that tends to spawn tons of off-topic threads, and inhibit the new people to write on the interesting topics. I fear that the only way to keep the anti-spam systems up to date, @Koni , would be to use some machine learning on your side too. So a technological progression escalates geometrically in that field, an thirty years from now we can hail Aleph as the shield against the Spam! And that, folks, is why independent, non-Aleph AI is so persecuted in the Human Sphere. The war reached a point so high only Pizza Hut survived, and Aleph concluded that having good and prosperous days required the total erradication of any other AI!!!! That, and the three seashells.
For over a decade email systems have been using text analisys to figure if a message matched valid emails or undesired ones. Sadly, support is not equal for all languages. As for smartness, you should check the issues "AI" have with traffic signals with stickers, crap or graffiti over them. Or the experiments to make them see something else. *giggle*
Very consistently early-morning in China, must be coded to look for Euro-US forums where the mods might be asleep. Still, at least it's not posting ponies.
How many mods here? All in same time zone? The staff page appears empty, so hard to find mods when in need. Euro-US sometimes also have Alaskan, Hawaiian, Australian or New Zelanders, and of course, people with weird schedules/shifts.
Wild idea, if user gets a small number of reports from different users of trusted background (tens or hundreds of valid posts, or like based, or time based, or combination, or special class "spam brigade"), account and related posts get into moderation. There is an unmaintained add-on that does something like that, but for individual posts, not accounts. Sad, sad state of anti spam tools by default. It's 2018... *snif* Edit: after some investigation, it seems warnings can trigger actions like ban for some time or change to special user group (still looking if hiding posts is possible or included via ban/new group), and that giving warnings is possible without getting more permissions. IOW, "spam brigade" looks factible based in public manual, support threads, etc.
My point was more related to how machine-learning development oriented to (limited) AI production used "anti-bot checks" as a measure for specific code developments. For example, how new code for visual processing allows a bot to circumvent the "are you a robot?" checkbox... XD (or, on a more specific way, the "select pictures with roads", "select pictures with street signs", etc...). I assume a standard evolution against bot-spamming will eventually imply the obfuscation of the true shapes behind camouflage specifically designed to spoof AI-controlled visual recognition (there is a "funny" video on how a 3D printed turtle is recognized as an automatic rifle by an AI in lab conditions...).
It's here again, this time in "presentaciones": https://forum.corvusbelli.com/forums/presentaciones.33/
The things I saw were signals with crap, not turtles, on streets. Which IMO is a more serious issue, it wasn't on purpose and the planned use is dangerous. Basically, it demostrated the problems of real life, not just "imagine if we could confuse the system" lab time spending (which I also have seen, including the "random image looks like..."). OTOH, at least one Western company offers "cheap humans" to solve problems, so humans could be doing the human part, and robots doing the "copy/paste"; similar things are going in Russia and Asia even if they don't offer the services to everyone. Just pick the best tool for each ocasion, meatbag or glassblock. There are other sources too, like unpaid humans (captcha for A is in reality one for B that has been passed around), or even ones hold hostage (instead of computer just mining coins, the human must also solve captchas or something bad happens). Also, Aleph will push the spam, if it makes more money than stopping it. Edit: another day, another Chinese pork-in-a-can truck load, with some Korean meat desserts.
Well, the possibilities are... interesting. In the chinese sense. Print a bunch of self-adhesive tags and paste them around the street: suddenly the cameras can't see some people. Or see plastic bags as weapons. Or children as enemy combatants... Not a single comforting idea, I know (can't see people is not bad? Say that with an auto-driven car... -.-U)
Another day, another rush. Koreans must be on holidays, but Chinese fill the slot perfectly. This time I went and played with some "AI"... gosh, it catches them fast, even when not sure, the guess was still correct. Maybe I should collect more samples. The first "AI" I tried time ago wasn't good with non Western languages, but this other doesn't care from what I read in tech paper. *lightbulb* I watched a video in which the driver tried to reproduce an accident. The silly car followed the white line without realizing it started to mean "you are out the road" when the highway split in two and the paint for one side was worn off a lot (think Y with bottom and one side stick white, the other side stick mostly grey). Driver braked, otherwise was going direct into Jersey barrier (or whatever that cement block was). Drive assist isn't very smart yet. Edit: nah, Korean just missed the bus and started late. Captured some extra samples of both kinds now.
Here it comes again... https://forum.corvusbelli.com/threads/mquq-2727586870rimqurimqurimqurimqumacquarie-university.19482/
I've sent a PM to @psychoticstorm with a solution proposal a few days ago. Hope they will implement it soon.