If this guy thinks the fact that so many Americans believe Saddam Hussein was involved in 9/11 is evidence that a new news site is needed, he's off to a bad start. They believe that because they want to, not because there are no reliable news sources available.
The fundamental problem is real though. News sites, including social news like reddit, are riddled with errors. Finding the truth often requires a real effort on the part of the reader, and for the most part I'm just too lazy/busy.
Maybe we could share the effort. It's just an idea, not very original, maybe bad but here it is anyway: PGP had the interesting concept of a "web of trust" (it was used in other places as well such as advogato.org). Maybe this concept could apply to news sources as well. E.g. my friend Bob is an expert on topic X, so I tend to trust him on said topic. I have a reddit-like website where I can see a selection of news "approved" by friends I trust instead of a website where a dozen of dumb high school kids (I'm actually talking about the reddit crowd here) I've never seen upvote/downvote articles. As it's a web of trust, I trust people who Bob trust as well (but a little bit less, maybe).
There is a traditional solution to this, in theory. The theory goes that a newspaper will meticulously edit and fact check, because at some point, they are bound to write about something I happen to know a lot about, and then if they are wrong, I will no longer trust them on other things. Unfortunately, the theory works a little too well: virtually all newspapers, if you read a few issues cover to cover, contain enough errors about things that you happen to know a lot about that you start mistrusting them. But there are a few publications for which traditional solution seems to work. I have rarely found any mistakes in The Economist, for instance, when they write about things I happen to know a lot about, which leads me to trust them more than most other news sources when it comes to things I don't know.
It's funny that you mention the economist, because I used to feel the same way about them, but then a few years ago they published some article about Google that was full of errors.
Clearly the old model isn't working though. I think the solution must be collaborative. I don't know if Thoof is it, but at least it's in the right direction.
I think anyone who is a general-purpose journalist rather than a domain expert is going to have this problem. Journalists don't actually understand the stuff they write about; no one could, probably; so instead they have a lot of hacks for sounding like they know what they're talking about.
Come to think of it, though, domain experts are often wrong too. Think about programming books, for example. I've now learned enough about several other fields to be fairly sure this problem is universal, except possibly in math. (Though even mathematicians have often been wrong in the sense of overlooking something that to later generations will seem obvious.)
Basically, you have to be a very critical reader. I doubt there's a "solution" beyond that. This is something the very smartest people struggle all their lives to get better at-- to take in all the stuff they're told, and from that figure out what's really going on.
Sometimes the problem has nothing to do with lack of domain knowledge. For example, before I went to MIT I saw a NYT article about students protesting the proposed removal of the pass/no credit system for the following freshman year. The article slanted it as if the students were selfishly interested in not getting grades. It doesn't take a domain expert to realize that if you're an MIT student now, you won't be a freshman next year, and could not be affected by the removal of pass/no credit. Students were objecting because they felt it would have a detrimental effect on others.
People give journalists such a bad rap... but no one asks any other profession to write definitively about lots of different subjects, often with only 24 hours to do it. Given the constraints, journalists do a pretty good job most of the time.
However, it is exactly this that makes general purpose journalism and newspapers a dying breed - as a big paper, you can't afford 1000 domain experts you might only use a few times a year. But bloggers are just this - domain experts who can distribute their content widely enough to be just popular enough.
There is some famous example to do with classification of simple finite groups where a big theorem had been proved over the course of many papers by various people over the years, and then when somebody tried to write a big monograph about it, they found all sorts of holes that took years to close.
Anyway, I looked at Thoof when it was posted on TC and I got an invite. It's a waste of time. It just requires way too much work from the regular user to be worthwhile and viable. Only redeeming quality was the endless scroll :).
Ian Clarke is such an arrogant guy. Back in the day when he was nothing but a developer, he was arrogant as hell, I can only imagine how he is now.
And he's arrogant, and produces nothing but commercial flops. He's a typical close minded geek type, he thinks he knows everything, yet is unable to see the big picture.
Moral of the story: It's not difficult to be polite. There is no excuse for rudeness, but it will come back to bite your ass, even 8 years later!
Btw oknotizie ( http://oknotizie.alice.it ), a reddit alike system my startup developed for the Italian market, already has the ability for users called "power users" (having enough total rank and enough rank accumulated in the last month) to replace the URL with one better suited (for example not linkjacked).
So how does a recommendation algorithm get around the fact that interesting doesn't equate to accurate? Oh wait it doesn't. Although I do enjoy the irony of the first publicity for this site that will find the TRUTH is complete bullshit.
Also the implication "we could win netflix we just don't want to" is not very credible.
the fact that this system uses bayes inference means it is doomed for failure. bayes requires large amount of training data. Most users want instant satisfaction. That could only mean failure.
