08 February 2010

Functioning the dark methods

The rise of artificial intelligence has produced an unanticipated but predictable corollary: artificial stupidity. This phenomenon has been much in evidence recently, in the form of robot spam comments (not to be confused with trolling, which is a separate issue I've dealt with in previous posts). These "spamments" (thanks to God Lizard for the term) are thinly-disguised ads; the "stupidity" comes from the fact that they seem to be generated without the slightest thought being given to making them effective.

For example, one specific posting on this blog, which is so old I doubt anyone has even looked at it for months, is constantly being bombarded with spam containing links to porn sites -- and I do mean bombarded -- there are at least one or two such comments a day. Here is the text of one of them (with the links removed):


Well, that's easy for him to say.

Brilliant advertising strategy, huh? I like to think my readers are on average more erudite than the general population, but as for how many of them are so fluent in Japanese as to be in the market for erotica in that language -- well, I'm not sure.

Here's the text of another "spamment", which the bot at least had enough "brains" to put on a recent post which was still being read by people (I rejected it in moderation, though):

Yes indeed, in some moments I can say that I jibe consent to with you, but you may be in the light of other options. to the article there is still a suspect as you did in the downgrade delivery of this beg [commercial link deleted]? I noticed the axiom you have in the offing not used. Or you functioning the dark methods of development of the resource. I suffer with a week and do necheg

I'm not sure what it means to jibe consent to with somebody, especially somebody who has an axiom in the offing and is functioning the dark methods of development, but I suspect that it's illegal in the Bible belt. Do necheg afterwards and you'd be looking at some serious prison time, you pervert.

Seriously, it looks like they're programming their bots with some basic grammatical structure and a dictionary of common words so that they generate sentences in the hopes of being mistaken for real human commenters, but the programs obviously still need some work.

Then there are the ones that just say "Great post! I would like to see more posts like that!" followed by the commercial link, which can fool you the first couple of times, unless they happen to appear as comments on a post about, say, a death in the family.

You'd think they would at least program these things to notice when a blog uses comment moderation, since that means that their inept advertisements will never appear, even momentarily. But I guess auto-posting comments is so cheap that it's not worth wasting any AI on such nuances.

Still, as long as I do have the moderation, spam isn't too much of a problem. It's a minor nuisance, something one can live with. Or, in the words of the persistent commenter on my ancient post:


Now who can argue with that?


Blogger Holte Ender said...

Never had "spamments" yet, plenty of trolling abusers, which got me so mad I didn't post for a week, which was probably their intention.

It must be so cheap and easy to send out those disguised messages for porn sites, or whatever they are trying to sell, that a one per cent return is a successful days work. The internet is like society, some people are trying to make it better, and they are hampered by people who are trying to make it worse.

08 February, 2010 06:27  
Blogger TomCat said...

I have mine set up so that comments on posts over two weeks old are moderated.

08 February, 2010 11:43  
Blogger godlizard (aka dotlizard) said...

I think you've hit the proverbial nail on the head with "it looks like they're programming their bots with some basic grammatical structure and a dictionary of common words so that they generate sentences in the hopes of being mistaken for real human commenters" -- I've not been able to dig up anyone advertising this explicitly, but I'm sure it exists on a level that's not publicly described as part of a SEO/SEM service, perhaps under the guise of "put your link on thousands of relevant forum threads and discussions".

What you've described is social engineering, and you've exactly charted the course of its progress, from the early "nice post!" to the more sophisticated modern bots, which attempt to give the impression of a thoughtful response -- they know that comments are a psychological reward for (many) bloggers, and use that against them. And there's no motivation to detect comment moderation, with bots it's just volume, it wastes only a fraction of a second to submit something that will go nowhere.

There's an even more insidious version out there, one that targets certain keywords and forms a generic comment on a particular topic, probably from randomized sentence fragments specific to that subject (and much more literate than your 'functioning the dark methods' bot, which is using a rather primitive algorithm that is little more than random phrases). The programmer who wrote that bot sucks at his or her job - the good ones are using Markov chains and Bayesian sampling. Spam is huge business, and it pays well. It helps to be ethically challenged, but being hungry and unemployed works too - lots of smart people go over to the "dark methods."

08 February, 2010 12:59  
Blogger Infidel753 said...

HE: That's what I'd figured -- it's so cheap that any refinements to make it more "targeted" would cost more than they'd be worth.

TC: That probably helps, although the majority of span comes to new posts -- I've rejected three spam comments on this very post about spam comments since I put it up!

GL: Thanks for the insights. You seem to know a lot more about the technical side of this stuff than I do, for which I congratulate you -- personally I'd like to flog some of these people with a Markov chain until they had to take a whole bottle of Bayesian aspirin.

Unfortunately, it sounds like the problem is just going to increase due to growing AI sophistication. Well, maybe one day the spambots will actually be able to generate comments as relevant, literate and insightful as real humans -- at which point I might even start letting them participate in the discussion. Of course by then they'd probably be starting their own blogs.

Spam is huge business, and it pays well. It helps to be ethically challenged, but being hungry and unemployed works too

Well, I'm not unemployed, but aside from that.....I wonder if they take moonlighters.....:-)

08 February, 2010 13:20  
Blogger Octopüß said...

Here are my experiences from Land-o-Spam. A Spam-Bot left a comment under one of my posts, Google v. The Great Firewall of China:

Anonymous said...
My friend and I were recently talking about the ubiquitousness of technology in our daily lives. Reading this post makes me think back to that discussion we had, and just how inseparable from electronics we have all become.

I don't mean this in a bad way, of course! Ethical concerns aside... I just hope that as the price of memory drops, the possibility of uploading our brains onto a digital medium becomes a true reality. It's one of the things I really wish I could see in my lifetime.

(Posted on Nintendo DS running [url=http://kwstar88.insanejournal.com/397.html]R4 SDHC[/url] DS FFBrows)

What I’ve noticed recently, spammers actually read these posts and leave inventive, almost charming, comments … except that they are spammers, and I really hate spammers. There are times when I am tempted to reply: “Instead of shit for brains, you have silicon for brains and a fetish for Nintendo that borders on bestiality.” But the Zone is supposed to be a place of dignity and decorum … (dammit, do we hafta?)!

Captain Fogg says, the delete button should emit an audible scream every time you press it. That would at least be sweet revenge.

08 February, 2010 19:36  
Blogger Pamela D. Hart said...

Infidel: Goodness! If I have to start dealing with "spamments" I’m really gonna lose my mind!

Not to mention, I don’t need to consent to any jibe considering I’m a married woman hence I wouldn’t even consider a downgrade since my current husband is a mighty sufficient axiom and is functioning quite well in the dark with methods I will not speak of in mixed company.

08 February, 2010 21:10  
Blogger godlizard (aka dotlizard) said...

Yes, the AI will make them harder and harder to spot -- a good rule of thumb (if you're not using the awesomeness of Akismet on WP, which rocks) is that no matter how relevant a comment, if the commenter leaves a link to a degree mill or a v*agra sale, the link itself is spam, and the comment attached to it is also spam. Even if they're using the low-tech method of paying a ten year old in Bombay fifty cents an hour to copy/paste with minor edits here & there, it's still spam. You can approve the comment if it's really super-interesting, just remove the link.

Those links are intended to scam google's page rank algorithm, and that's just bad. It means the bad guys win, and people who work very hard to create worthy content and get it noticed through hard work in legitimate channels, suffer.

Yeah, I'm kinda into this.

09 February, 2010 00:52  
Blogger magpie said...

I'm fluent in Japanese.

What's written in your post there has not rendered the kanji though, so all I can see are the phonetic hiragana. You wouldn't want a translation anyway.

I hasten to add too that this sort of spam is not my fault. I have never had it appear in my blog or inbox.

09 February, 2010 04:19  
Blogger Infidel753 said...

Octopüß: I had that exact same "comment" myself once, word-for-word. I guess this is another approach -- write something so generic that it could sound like a plausible comment on a wide range of posts. That's pretty much what astrologers and cold readers do.

Dignity and decorum have their place, but are useless when not reciprocated. A screaming (or machine-gun sound-effect) reject button is a feature which Blogger should consider atting to its moderation screen.

Pamela: Hi there! Spamments, unfortunately, seem to be the fate to which all bloggers ultimately come. The question is how we face it.

I figured you for the kind of person who doesn't take no jibe from nobody, and it sounds like things are well in hand. The Japanese spamments are actually racier (and more coherently so), but on an American blog, what's the point?

GL: I've actually done that on a very few occasions (let the spamment post but without the link), mainly when it was worth making fun of. Interesting point about fooling the Google algorithm, though. It's tough for bloggers even with good content to find an audience, and I wouldn't want to contribute to that.

Thanks for visiting, all.

09 February, 2010 04:19  
Blogger Infidel753 said...

Magpie: Japanese text displays inconsistently on Western computer systems. On my work computer I can see it normally, kanji and kana both -- on my home computer all that appears is a series of squares. I've never seen a computer that displays kana but not kanji, but I can't say I'm surprised.

I'm not fluent in Japanese, but I can read enough to know what it's talking about.

09 February, 2010 04:40  
Blogger Elizabeth said...

I suffer with a week and do necheg

Infidel, you may laugh, but I jibe a consent to this anonymous poster.

This is indeed a serious and much underreported problem among people of a certain age. I've been suffering with a week for several years now, and only after I started practicing necheg, my symptoms lessened enough to start functioning the dark methods.

If my progress -- thanks to necheg, of course -- continues at the same rate, I may even be able to downgrade delivery of this beg, which is no small feat, to be sure.

P.S. I too have noticed the axiom you have in the offing.

09 February, 2010 23:49  
Blogger Infidel753 said...

I've been suffering with a week for several years now, and only after I started practicing necheg, my symptoms lessened

This seems to be a more common health issue than I realized. Perhaps we should lobby Congress to add an amentment to the health bill reducing the length of the week, or at least subsidizing necheg. The teabaggers would, of course, downgrade to even darker methods, but it's pretty much axiomatic that they would do that.

10 February, 2010 04:32  
Anonymous Hugo Grinebiter said...

I defer to Godlizard on techie stuff, but to me the axiom-offing post looks much more like something written in another language by an actual human and making sense in that language, but then run through Babelfish or similar.

There are sites for genuine advertising copy and hotel notices from East Asian sources that are not much better. I have read Japanese auto patents, and been assured that other Japanese can understand what we can't. They are surely translating back into Japanese to recover the original meaning -- I have to do that myself sometimes to make sense of what Scandis are trying to say in English.

Some of the Asian English productions are delightful. I love the one about hotel ironing services: "flattening your underwear with pleasure is the job of the chambermaid". Now you see, that's linguistically more or less correct, semantically not so much. Or consider this, from Copenhagen Airport, an advert for a baggage-handler: "We take your bags and send them in all directions". Try to explain to a non-native-English-speaker exactly why that is wrong; a baggage handler should send bags in all directions, after all.

The "beat poetry" I get is something different. Here's one that came in since I emptied the bit bucket (Akismet) this morning:

"Lights dancing stability sodium alendronate and suddenly use of macrobid vs cipro water breathing taking synthroid bontril cockatrice still steroids mechanism of action than trolls" (and so on for about 20 times this length). Now this is untouched by human hands.

19 February, 2010 12:26  
Blogger Infidel753 said...

Unassisted translation from Asian languages to English can be problematical. Japanese in particular has an entirely different kind of syntax than Western languages or even Chinese have.

(I have made embarrassing mistakes when trying to communicate in Japanese too. It works both ways.)

That "beat poetry" looks like a purely random string of words to me. Serious efforts to translate from another language to English sometimes at least generate entertainment by accidentally sounding like sexual innuendo -- and I'm sure our efforts to use other languages we don't know well sometimes produce the same result.

19 February, 2010 13:52  

Post a Comment

Links to this post:

Create a Link

<< Home