1. The discoverer pitches the claim directly to the media.
2. The discoverer claims that a powerful establishment is trying to suppress their work.
3. The scientific effect described is always at the very limit of detection.
4. Evidence for the discovery is anecdotal.
5. The discoverer says a belief is credible because it has endured for centuries.
6. The discoverer has worked in isolation.
7. The discoverer must propose new laws of nature to explain an observation.
All very wise. I wonder if it would be possible to elaborate a similar set of markers for the reliability of web rumours? After all, discussion forums, blogs, news sites and the like are running over with extremists who validate their crazed rants by referring to what they claim are authoritative sources. The point being that very few people ever follow the link or wonder where the astonishing news came from. If there's anything that bears this out, it was the "Kerry/Fonda photo" affair - the faked photo was prepared by a bunch of far-right nutters who posted it all over some high-traffic forums. Then it got onto a low quality US news site (NewsMax), and many more linked to their story. Then it burst and hit the mainstream media, oddly enough after it had already been debunked by Newsday. I'm thinking of similarities between dodgy sources, things like the weird way US fringe news sites like NewsMax and WorldNetDaily (another favourite for gun clubbers and conspiracy odds) advertise quack remedies heavily as well as get-rich-quick schemes and the like. Perhaps spam filtering might offer an analogy? (Strangely enough, my stats provider seems to be going that way. I've noticed that it tries on occasion to run a dialler, and today's amusing popup was "Are YOU Stupid? On-Line Personal IQ tests!")
Come to think of it, here is a provisional version for discussion:
1. If the source has dodgy advertising or dodgy web practices, assume the content is dodgy too.
2. Credibility is cumulative - if the rest of its output is crazed, chances are the bit you're looking at is unreliable as well.
3. Facts and comment.
4. Use the Google principle - links are votes. Cranks and propagandists tend to band together for mutual support.