I've read quite a number of blog posts recently that have been written in English that's seemed strangely clunky and unnatural, in a style very similar to the intro text on the Datapresser home page.
I initially put it down to posts being authored by people for whom English is a second language, but they're US-based blogs. This initially dodgy-sounding idea starts to seem plausible.
I haven't read that blog before, but now scrolling down through some recent entries, it's all very clunky English. On the other hand, the person writing it is named Ernst-Jan Pfauth.
So the game of the week must be 'Which mainstream bloggers are using content generation tools?'
And the next big thing in startups is going to be filtering out algorithmically generated content.
Thus leading to a coevolutionary arms race where generators and filters are getting better and better until mere humans are outclassed. Automated comment bots and semantic maps of key demographic groups become dominant, generations of marketing lore gets coded into verbiage generators leading to an end state where the entire universe of online discourse is generated by Robert Scoble pushing a button, repeatedly.
The personal subscription $25CDN is "limited" to 500 posts/day. Even TC doesn't post that much.
I think the real discussion is moving to Twitter-like mediums (short and quick - who has time for blogging). Looking at Techmeme, it looks like a lot of the discussion isn't really - just rehashing (no time for analysis)
Their volume of articles really does not seem implausible given their number of writers and the abundace of information. Sure, they put out quite a few articles a day, but if that is their mission, it is not surprising. What they do is far from inhuman.
That said, this product seems like it could be of use for them.
Actually, datapresser sounds like a very interesting side-project to work on, involving natural language processing and machine learning. I'm pretty impressed, sounds like it could have been a side project turned into a viable startup business.
I'm calling BS on that claim. If the software could do anything more advanced than replacing random words from a thesaurus, there would be a lot of applications for it more lucrative than populating spamblogs.
This is really really awful for the future of the Internet. Generating random content is specifically the kind of garbage that we don't want on the web. It's the definition of 'free-ridership' -- lets create a bunch of valueless nonsense programmatically, and massively destroy true value (real useful information) and then PROFIT.
As general content degrades, all the more that eventually we're going to see some form of information and content delivery certification emerge, not unlike what Underwriters Laboratories do for for product and material goods safety and quality, or what the USDA and FDA try to do for food and drug quality. Perhaps something akin to the movie and game content ratings, but adding fact checking.
A voluntary certification system could help people trust content. Search engines could apply web content providers' quality group affiliation (or lack thereof) to search heuristics.
Maybe it will be NGO based, maybe government run, depending on where you are in the world. I'd opt for the NGO approach myself, with a healthy number of competing organizations, especially considering China's effective use of content control as an extreme case.
A danger here, though, is if we move to "seals of approval" web content and the majority of individuals and groups avoid content that does not have a "seal of approval" then there are going to be barriers to entry, such as the various costs that members of associations pay for services.
There's a Philip K. Dick story that involves a machine like this that is used to generate political speeches. Can't quite remember the title. Anyone know?
This is a stupid joke. Nobody uses this type of software for real blog posting. You might say it really doesn't even pollute the web because content generated with these things probably aren't going to rank very well and most people won't see them. The real purpose is to create a massive number of links which can be used for supporting the rank of a more legit site. So this is Google's problem to deal with.
This reminds me of SCIgen: http://pdos.csail.mit.edu/scigen/
These guys entered their auto-generated research paper to a conference and actually got invited to present it at a seminar.
Using such a tool sounds like a vast lie from the companies benefiting from it. Isn't impersonation a crime? It would be fun to sue for massive amounts one of the fake users and then the company that generated would be liable:))) If ever proven so through.
It might sound surprising but CNet also uses similar techniques to generate huge parts of their product reviews.
The technique is the biggest threat to the current crop of 'content' driven search engines (read Google and everyone else) and has the potential to wreak havoc with current ranking systems. Given the lucrative nature of the automatic content generation business, this is also inevitable.
27 comments
[ 4.6 ms ] story [ 74.6 ms ] threadI initially put it down to posts being authored by people for whom English is a second language, but they're US-based blogs. This initially dodgy-sounding idea starts to seem plausible.
And the next big thing in startups is going to be filtering out algorithmically generated content.
Thus leading to a coevolutionary arms race where generators and filters are getting better and better until mere humans are outclassed. Automated comment bots and semantic maps of key demographic groups become dominant, generations of marketing lore gets coded into verbiage generators leading to an end state where the entire universe of online discourse is generated by Robert Scoble pushing a button, repeatedly.
I think the real discussion is moving to Twitter-like mediums (short and quick - who has time for blogging). Looking at Techmeme, it looks like a lot of the discussion isn't really - just rehashing (no time for analysis)
That said, this product seems like it could be of use for them.
I'm not sure whether or not I hope it's a joke.
-1 feeds unhealthy TC obsession
-1 weak attempt at humor
-1 single joke stretched out too long
-1 ungrammatical headline
-1 awkward writing in article
In the News.YC of my dreams, everyone who upvoted this article would be disenfranchised.
Weak.
A voluntary certification system could help people trust content. Search engines could apply web content providers' quality group affiliation (or lack thereof) to search heuristics.
Maybe it will be NGO based, maybe government run, depending on where you are in the world. I'd opt for the NGO approach myself, with a healthy number of competing organizations, especially considering China's effective use of content control as an extreme case.
A danger here, though, is if we move to "seals of approval" web content and the majority of individuals and groups avoid content that does not have a "seal of approval" then there are going to be barriers to entry, such as the various costs that members of associations pay for services.
Also works on YouTube.
The technique is the biggest threat to the current crop of 'content' driven search engines (read Google and everyone else) and has the potential to wreak havoc with current ranking systems. Given the lucrative nature of the automatic content generation business, this is also inevitable.