Tim Bray wrote a long column about analysts awhile back, and I finally have time to comment on it, seeing as how I'm stuck at 30,000 feet with a bunch of flagged items in BottomFeeder. Here's an anecdote Tim used to jump into the post:

The perception of corruption, whether it’s true or not, hasn’t gone away. Shortly after I joined Sun, I was trying to figure out why large parts of the industry seemed hell-bent on re-creating CORBA, only more complex and less efficient, under the WS-* banner. One senior technology strategist, not from Sun, told me “Obvious! It’s because IBM and Microsoft paid the analysts megabucks, megabucks I tell you , to go out and tell everyone that this was where the future was, and anyone who wasn’t going that way was dog meat.” Mind you, he’d had a few beers. But that’s not the only time I’ve heard that particular suggestion.

I have to agree with Tim's quoting of an old standby for this: "Never ascribe to malice that which can be explained by stupidity". Industry analysts aren't evil, and they aren't all on the take. I'm sure some are; there are bad apples in every field - and trade analysis is hardly immune. That's not the main problem though. The analysts are looking for the same thing that IT shops and vendors are looking for: The next big thing (tm). IT managers want to know what it is, because there's safety in numbers; if "everyone" is using a certain technology, then the thinking is that support will be easy to find, and it can't be too bad. Analysts want to be ahead of the curve, so they are scanning for that "big thing" - because once they find it, they have something to point to.

What that leads to isn't corruption, but herd behavior. Without consciously working at it, IT managers and analysts reinforce their own preconceived notions, and are prone to jumping on bandwagons. Witness the rise of OO as a buzzword, relational databases before that, and Java and WS* more recently. It's not that these were necessarily bad, even - they just became reinforcing memes that spread across the tech landscape. So what you end up with is a system that can easily be construed as corrupt - the big analysts are pushing the big technologies that the big vendors want to push. I think the actual behavior is more subtle than that, but - as Tim says - it's certainly a perception problem for the analyst firms:

It’s a problem for the analysts, and their customers, and for the industry, that there’s this elephant in the room. Because I totally believe that we need analysts. I know for a fact that there are those who read people like me and Don Box and Bob Sutor and use what we say about Java or messaging stacks or ODF as serious business input. But dammit, we’re vendors, our paychecks depend on selling you expensive stuff! At least with us, our hearts are on our sleeves and the conflict of interest is screamingly obvious.

It seems that in a rational world, there’d be a place for professional intermediaries; someone who has a non-tech business to run doesn’t really have time to drill down on whether crazies like me who are dissing WS-* are right or wrong. They should be able to outsource that research. (By the way, no analyst from a mainstream firm has ever raised the WS-* issue with me, which seems a little weird).

I think analysts are going to have to do what Tim suggested - they are going to have to disclose, using journalist style ethical standards, what vendors are paying them, and how much they are paying them. As Tim said, with us vendors the bias is obvious - don't come to me, for instance, if you are looking for an objective opinion about Java (or Smalltalk, for that matter). The analysts, on the other hand, are striving to be that objective voice. They need to make sure that the audience can make a judgement call based on all the inputs.