4 ms·

“Do you guys even CARE what I write? Does it make any difference if it’s good or bad?” I said. “Not really,” was the reply. If he had any programming backgrou

“Do you guys even CARE what I write? Does it make any difference if it’s good or bad?” I said. “Not really,” was the reply. If he had any programming background, he missed a golden opportunity to write a markov text generator which would have let him meet his deadlines without the stress.

spaghetti 15y ago
You beat me to it :-). I'm curious to what extent this is being done today.
- ChuckMcM 15y ago
  In the SEO world you come across entire forum sites of imaginary users all talking about $SEOTarget product and how create $TargetWebSite is all amongst their little markov robot selves. The goal is to create links which rank highly and raise the target in search results for search engines.
jerf 15y ago
Obviously a straight-up Markov Chain generator isn't going to work, but a smarter system that would allow you to basically sketch the article and then have the system automatically babble in English to fill out the word count while not actually needing any additional information strikes me as feasible, though not trivial. Somebody's doing something similar for sports reporting: http://thenextweb.com/media/2011/04/18/robot-journalist-writes-a-better-story-than-human-sports-reporter/ http://thenextweb.com/media/2011/04/18/robot-journalist-writ... You'd have to tweak it for a different domain but I'm sure that would be feasible for your average gossip-rag type story.
- hugh3 15y ago
  The Motley Fool also uses automatically-generated articles. For instance if you google the phrase "Being able to retire rich, or at least comfortable, is the goal of almost any investor" you'll find hundreds of articles from fool.com, each posing as an analysis of a particular company, and all generated with an algorithm which is fairly obvious after reading a couple of 'em. They're not completely worthless, though. Real data + automated analysis is far better than no data or no analysis.
  - orborde 15y ago
    That is disturbingly brilliant.
  - wging 15y ago
    OUCH. No mention of the contraction of the newspaper industry when discussing Gannett? >Next, we want to ensure that Gannett's stock has the ability to rise over the next five, 10, or 20 years. A company that's growing its net income has the best possible chance to see its share price rise over time. Of course, we can't predict the future, but we can look back to get an idea of how the company has performed in the past in order to try to ensure future earnings growth. Over the past five years, Gannett has shrunk its net income at an annual rate of 13.9%. Unfortunately, Gannett has run into its own share of problems, and the financial collapse of 2008 certainly couldn't have helped either. So the company has been unable to grow earnings, which doesn't exactly mean that it won't in the future, but it's certainly not the greatest of signs. from http://www.fool.com/retirement/general/2011/04/26/should-you-retire-with-gannett.aspxhttp://www.fool.com/retirement/general/2011/04/26/should-you-retire-with-gannett.aspx http://www.fool.com/retirement/general/2011/04/26/should-you... It's kind of clear that it's auto-generated after just one. The only other possibility would be that "Jordan DiPetro" had a real point to make about what stock stats can tell you without any real knowledge...
  - knowtheory 15y ago
    My first reaction was disbelief, "what? the Motley Fool doing something that is clearly unethical?" Then i googled as you mentioned, and indeed found all of these identically phrased articles. But upon reading them i'm impressed at exactly how generic this template is. I'm not sure they're actually doing harm, because frankly, these articles aren't analysis. They don't actually offer any judgement about the stocks in question. They're basically a textual presentation of some stock data that is designed to serve as a feeder into fool.com. I can't say that i'm okay with it, but it's not as catastrophic as i had initially imagined. Still definitely SEO spam though :\ I am disappoint.
beambot 15y ago
Ha! Reminds me of the Automatic CS Paper Generator [1] from MIT. [1] http://pdos.csail.mit.edu/scigen/ http://pdos.csail.mit.edu/scigen/

spaghetti 15y ago

You beat me to it :-). I'm curious to what extent this is being done today.

ChuckMcM 15y ago
In the SEO world you come across entire forum sites of imaginary users all talking about $SEOTarget product and how create $TargetWebSite is all amongst their little markov robot selves. The goal is to create links which rank highly and raise the target in search results for search engines.

jerf 15y ago

Obviously a straight-up Markov Chain generator isn't going to work, but a smarter system that would allow you to basically sketch the article and then have the system automatically babble in English to fill out the word count while not actually needing any additional information strikes me as feasible, though not trivial. Somebody's doing something similar for sports reporting: http://thenextweb.com/media/2011/04/18/robot-journalist-writes-a-better-story-than-human-sports-reporter/ http://thenextweb.com/media/2011/04/18/robot-journalist-writ... You'd have to tweak it for a different domain but I'm sure that would be feasible for your average gossip-rag type story.

hugh3 15y ago
The Motley Fool also uses automatically-generated articles. For instance if you google the phrase "Being able to retire rich, or at least comfortable, is the goal of almost any investor" you'll find hundreds of articles from fool.com, each posing as an analysis of a particular company, and all generated with an algorithm which is fairly obvious after reading a couple of 'em. They're not completely worthless, though. Real data + automated analysis is far better than no data or no analysis.
- orborde 15y ago
  That is disturbingly brilliant.
- wging 15y ago
  OUCH. No mention of the contraction of the newspaper industry when discussing Gannett? >Next, we want to ensure that Gannett's stock has the ability to rise over the next five, 10, or 20 years. A company that's growing its net income has the best possible chance to see its share price rise over time. Of course, we can't predict the future, but we can look back to get an idea of how the company has performed in the past in order to try to ensure future earnings growth. Over the past five years, Gannett has shrunk its net income at an annual rate of 13.9%. Unfortunately, Gannett has run into its own share of problems, and the financial collapse of 2008 certainly couldn't have helped either. So the company has been unable to grow earnings, which doesn't exactly mean that it won't in the future, but it's certainly not the greatest of signs. from http://www.fool.com/retirement/general/2011/04/26/should-you-retire-with-gannett.aspxhttp://www.fool.com/retirement/general/2011/04/26/should-you-retire-with-gannett.aspx http://www.fool.com/retirement/general/2011/04/26/should-you... It's kind of clear that it's auto-generated after just one. The only other possibility would be that "Jordan DiPetro" had a real point to make about what stock stats can tell you without any real knowledge...
- knowtheory 15y ago
  My first reaction was disbelief, "what? the Motley Fool doing something that is clearly unethical?" Then i googled as you mentioned, and indeed found all of these identically phrased articles. But upon reading them i'm impressed at exactly how generic this template is. I'm not sure they're actually doing harm, because frankly, these articles aren't analysis. They don't actually offer any judgement about the stocks in question. They're basically a textual presentation of some stock data that is designed to serve as a feeder into fool.com. I can't say that i'm okay with it, but it's not as catastrophic as i had initially imagined. Still definitely SEO spam though :\ I am disappoint.

beambot 15y ago

Ha! Reminds me of the Automatic CS Paper Generator [1] from MIT. [1] http://pdos.csail.mit.edu/scigen/ http://pdos.csail.mit.edu/scigen/