# can a spambot produce this almost-reasonable text? **URL:** \ **Category:** Factual Questions\ **Created:** [March 12, 2013, 6:18am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685 "2013-03-12T06:18:07Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![chappachula](https://avatars.discourse-cdn.com/v4/letter/c/d2c977/32.png) [@chappachula](https://boards.straightdope.com/u/chappachula)\ **Post date:** [March 12, 2013, 6:18am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/1 "2013-03-12T06:18:07Z")
In [this IMHO thread about legalizing marijuana](http://boards.straightdope.com/sdmb/showthread.php?t=684857), there is a post (#8) from a new guest, making his first post: I’ve copied the entire post: > [@](#): > > •-SMILE instead of TRIAL-• > To view negative response to marijuana is basically all very uneducated of its real reason existing. > Never even 1 time has it done what alcohol does. It was actually LEGAL prior to alcohol prohibition. One of reasoning madhe illegal was black persons mainly had it and black life was more back of the bus. I am amazed never spoken is it was a creation by God and medicinally used and still is. > Never in any sermons do I hear it spoken of. Read it again completely now. > In Colorado & Washington crime has lowered and police have more time to spend on important things. I worked at a club for 5 years and plenty to say. My question is: is this a real person, and if not, how does the robot work? I suppose the post could be simply a teenager who is totally stoned, but still somehow coherent enough to read the instructions to register at this site,before typing an incoherent reply. But more likely, the post is from a bot of some sort. But it doesn’t appear to have any financial incentive–no advertizing, no links to anything. So why would somebody make a bot that : 1) doesn’t work very well, and 2) doesn’t earn money? Some of the language could be a robot that samples texts and glues phrases together, but on the other hand,there is a certain amount of, not exactly logic, but “connectedness” within each paragraph.Is this the level of artificial intelligence these days? So—my first question is : whodunnit? But my second -and main-question is: (asssuming it’s a bot)–WHY? ---
**Author:** ![Gary\_Robson](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/gary_robson/32/3448_2.png) [@Gary\_Robson](https://boards.straightdope.com/u/Gary_Robson)\ **Post date:** [March 12, 2013, 6:25am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/2 "2013-03-12T06:25:02Z")
Remember that we have members for whom English is a second (or third…) language. I don’t know if that’s what we’re looking at here – it’s late and I’m not going to stay up and do detective work – but that’s certainly a possibility. ---
**Author:** ![davidm](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/davidm/32/225_2.png) [@davidm](https://boards.straightdope.com/u/davidm)\ **Post date:** [March 12, 2013, 6:51am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/3 "2013-03-12T06:51:28Z")
As far as financial incentive, there are bots that will attempt to establish credibility over time by posting a few seemingly relevant posts spread out over a couple of days before creating the money post. That said, I think the non-native speaker hypothesis is more believable. A bot determining that a thread is about legalizing marijuana and then creating an appropriate message would be pretty damned sophisticated. ---
**Author:** ![tellyworth](https://avatars.discourse-cdn.com/v4/letter/t/977dab/32.png) [@tellyworth](https://boards.straightdope.com/u/tellyworth)\ **Post date:** [March 12, 2013, 7:09am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/4 "2013-03-12T07:09:33Z")
> [@chappachula](#): > > My question is: is this a real person, and if not, how does the robot work? > I suppose the post could be simply a teenager who is totally stoned, but still somehow coherent enough to read the instructions to register at this site,before typing an incoherent reply. > … > So—my first question is : whodunnit? > But my second -and main-question is: (asssuming it’s a bot)–WHY? The text is generated by a bot. It consists of sentences taken from a large corpus, cut up and randomly reassembled. There is no intelligence to it. Some part of the process of creating the account and posting the text may have involved manual intervention, I don’t have enough information to say in this case. As for why, that depends what you mean. Why is it randomized in this particular way? To try to defeat the spam filters commonly in use on forums. Why does it not contain an obvious link or purpose? To seem plausible enough to pass by an inattentive moderator. What are they trying to achieve? They’ll come back later and post advertising links, because their client paid for traffic and/or search engine rankings. ---
**Author:** ![tellyworth](https://avatars.discourse-cdn.com/v4/letter/t/977dab/32.png) [@tellyworth](https://boards.straightdope.com/u/tellyworth)\ **Post date:** [March 12, 2013, 7:11am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/5 "2013-03-12T07:11:54Z")
Oh, as for the apparent relevance to the thread: they’re using a corpus of text loosely related to some topic they’re advertising. And they’re seeking out forum threads that rank well for keywords related to that topic. ---
**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\ **Post date:** [March 12, 2013, 7:37am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/6 "2013-03-12T07:37:15Z")
> [@tellyworth](#): > > The text is generated by a bot. It consists of sentences taken from a large corpus, cut up and randomly reassembled. There is no intelligence to it. I know nothing of such things but know that you have expertise. Nevertheless, I’m surprised. Surely software smart enough to combine vaguely related phrases to each other could have been smart enough to avoid simple grammar errors. Consider “I worked at a club for 5 years and plenty to say.” But perhaps, since the overall content of computer-generated text is doomed to appear low-intellect, the low-intellect grammar is _deliberate_: good grammar would have been incongruous. ---
**Author:** ![Simplicio](https://avatars.discourse-cdn.com/v4/letter/s/c37758/32.png) [@Simplicio](https://boards.straightdope.com/u/Simplicio)\ **Post date:** [March 12, 2013, 7:40am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/7 "2013-03-12T07:40:10Z")
I dunno, wouldn’t the individual portions be grammatical, or at least have better spelling if that were the case. I’m sticking with it being a combination of a real person who is both a non-native speaker and baked out of their gourd. ---
**Author:** ![Simplicio](https://avatars.discourse-cdn.com/v4/letter/s/c37758/32.png) [@Simplicio](https://boards.straightdope.com/u/Simplicio)\ **Post date:** [March 12, 2013, 7:46am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/8 "2013-03-12T07:46:40Z")
There’s a user on [Youtube](http://www.youtube.com/user/jjerel1953) with the same name, similarly clipped grammer, and at least one post on marijuana legalization. Granted I just skimmed, but so far as I can tell, they’ve had an account there for several years and yet haven’t tried to sell anything. So I vote not a bot. Maybe **Tellyworth** is the replicant!! ---
**Author:** ![tellyworth](https://avatars.discourse-cdn.com/v4/letter/t/977dab/32.png) [@tellyworth](https://boards.straightdope.com/u/tellyworth)\ **Post date:** [March 12, 2013, 7:51am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/9 "2013-03-12T07:51:51Z")
On further study: I misinterpreted something initially. There are indeed spammers that operate in that manner, and that produce text similar to this, but I this might not be one of them. Grammatical and spelling errors come from the source material, and from mashing sentence fragments together. ---
**Author:** ![tellyworth](https://avatars.discourse-cdn.com/v4/letter/t/977dab/32.png) [@tellyworth](https://boards.straightdope.com/u/tellyworth)\ **Post date:** [March 12, 2013, 8:08am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/10 "2013-03-12T08:08:29Z")
**Simplicio** is correct, this is a real person. My apologies to our guest. ---
**Author:** ![chappachula](https://avatars.discourse-cdn.com/v4/letter/c/d2c977/32.png) [@chappachula](https://boards.straightdope.com/u/chappachula)\ **Post date:** [March 12, 2013, 8:16am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/11 "2013-03-12T08:16:34Z")
Okay,… we seem to have a debate going here: either it’s a bot, or it’s a guy who’s baked out of his gourd. But maybe we’ve discovered something really profound, which scientists have been hoping to find for half a century: Because, if nobody can tell the difference between man and machine, isn’t that the perfect solution to the Turing Test? 🙂 ---
**Author:** ![chappachula](https://avatars.discourse-cdn.com/v4/letter/c/d2c977/32.png) [@chappachula](https://boards.straightdope.com/u/chappachula)\ **Post date:** [March 12, 2013, 8:27am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/12 "2013-03-12T08:27:06Z")
Or maybe we’ve uncovered a real-life conspiracy theory: the post at youtube which Simplico linked to could be from a bot–the same bot who just joined us. 🙂 (and to our new guest—if you are real, welcome to the SDope…But , please, try to write more coherently. Or else share some of that stuff you’re smokin’ --it’s been a long,long time since I was a student in the Sixties 🙂 ) ---
**Author:** ![Blake](https://avatars.discourse-cdn.com/v4/letter/b/a9a28c/32.png) [@Blake](https://boards.straightdope.com/u/Blake)\ **Post date:** [March 12, 2013, 9:01am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/13 "2013-03-12T09:01:13Z")
> [@tellyworth](#): > > Grammatical and spelling errors come from the source material, and from mashing sentence fragments together. Some of them also obviously derive from the use of thesaurus synonyms, presumably to prevent the text exactly matching the plagiarised source. This is why I also assumed this was a Bot. Had it been Bot generated, the first sentence would have started life as: "The observed negative response to marijuana is essentially highly ignorant of its real _raison d’être_. After a quick run through a thesaurus, the Bot produces “To view negative response to marijuana is basically all very uneducated of its real reason existing.” I’ve seen similar passages in genuine Bot generated messages, and every time it’s quite clear that words have been changed to Thesaurus synonyms that aren’t actually synonymous in context. So the passage seems to have meaning, but takes a lot of effort to decipher. ---
**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\ **Post date:** [March 12, 2013, 10:19am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/14 "2013-03-12T10:19:09Z")
> [@chappachula](#): > > But maybe we’ve discovered something really profound, which scientists have been hoping to find for half a century: > Because, if nobody can tell the difference between man and machine, isn’t that the perfect solution to the Turing Test? 🙂 We’re exploring a different Test: Can a sentient human imitate a text-generation bot? This one seems to have failed: grammar too bad, and text almost makes sense. ---
**Author:** ![tellyworth](https://avatars.discourse-cdn.com/v4/letter/t/977dab/32.png) [@tellyworth](https://boards.straightdope.com/u/tellyworth)\ **Post date:** [March 12, 2013, 10:35am UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/15 "2013-03-12T10:35:12Z")
> [@Blake](#): > > Some of them also obviously derive from the use of thesaurus synonyms, presumably to prevent the text exactly matching the plagiarised source. Content spinning. There’s a different method, peculiar to a prolific forum spambot, that produces content with more of a Horse\_ebooks vibe. That’s the one I was thinking of. **chappachula** , that’s not as unlikely as you might think - there are spammers who create fake identities, complete with profiles and bios on social sites. Expensive. And not the case here. Spammers make an effort to impersonate real humans. And sometimes humans are remarkably good at looking like spambots. ---
**Author:** ![Chronos](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/chronos/32/134_2.png) [@Chronos](https://boards.straightdope.com/u/Chronos)\ **Post date:** [March 12, 2013, 12:16pm UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/16 "2013-03-12T12:16:17Z")
It would be easier for a human to generate text like that directly than it would be to create a bot capable of creating it. And spambots that seek to establish “credibility” usually go for more generality, statements like “This is a very interesting topic” that could fit into any thread. ---
**Author:** ![moriah](https://avatars.discourse-cdn.com/v4/letter/m/b2d939/32.png) [@moriah](https://boards.straightdope.com/u/moriah)\ **Post date:** [March 12, 2013, 12:32pm UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/17 "2013-03-12T12:32:08Z")
> [@Simplicio](#): > > C’nest pas une bot You young people with your French slang! **[Ceci n’est pas une bot](https://en.wikipedia.org/wiki/The_Treachery_of_Images)** ---
**Author:** ![Doug\_K](https://avatars.discourse-cdn.com/v4/letter/d/b5ac83/32.png) [@Doug\_K](https://boards.straightdope.com/u/Doug_K)\ **Post date:** [March 12, 2013, 1:37pm UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/18 "2013-03-12T13:37:39Z")
> [@Chronos](#): > > It would be easier for a human to generate text like that directly than it would be to create a bot capable of creating it. And spambots that seek to establish “credibility” usually go for more generality, statements like “This is a very interesting topic” that could fit into any thread. It’s not horribly difficult. Scott Pakin’s [automatic complaint-letter generator](http://www.pakin.org/complaint) can generate things like this from just 5 keywords (6 if you count selecting male or female). [Complaint](http://www.pakin.org/complaint?title=Lord&firstname=Chronos&middlename=Q.&lastname=Doper&suffix=%2C+Esq.&gender=m&shorttype=f&pgraphs=3) (note: If you go to this link multiple times, you’ll get a different “complaint” letter each time.) ---
**Author:** ![dstarfire](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/dstarfire/32/5762_2.png) [@dstarfire](https://boards.straightdope.com/u/dstarfire)\ **Post date:** [March 12, 2013, 7:12pm UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/19 "2013-03-12T19:12:21Z")
It looks very much like what you’d get from one of those free online translation sites. ---
**Author:** ![Mister\_Rik](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/mister_rik/32/491_2.png) [@Mister\_Rik](https://boards.straightdope.com/u/Mister_Rik)\ **Post date:** [March 12, 2013, 8:41pm UTC](https://boards.straightdope.com/t/can-a-spambot-produce-this-almost-reasonable-text/652685/20 "2013-03-12T20:41:57Z")
> [@Blake](#): > > I’ve seen similar passages in genuine Bot generated messages, and every time it’s quite clear that words have been changed to Thesaurus synonyms that aren’t actually synonymous in context. So the passage seems to have meaning, but takes a lot of effort to decipher. > [@dstarfire](#): > > It looks very much like what you’d get from one of those free online translation sites. And it’s fairly common with somebody speaking English as a second language. My favorite real-life example was a Mexican man who was browsing the displays where I worked years ago. When I asked, “May I help you?” he replied, “I am seeing only.” It was clear that he meant, "I’m just looking’, but didn’t understand English well enough to catch the difference in meaning between “looking” and “seeing”.