I think such a business may succeed. Your biggest challenge will be what I'll call the "verification problem": How can a customer verify that you've provided them with a "good" product for their money?

Consider selling programming services. The client can request an application or website with certain functionality, and you use your superior expertise to create that functionality. Even though the client can't code, their experience from using the application/website can tell them that you delivered the goods.

Knowledge that is used to support strategic business decisions is not so easy to verify, mainly because it doesn't have consequences that can be directly experienced by non-experts.

So how do people get around this? Large consulting firms (McKinsey, BCG, Bain, etc.) spend considerable resources building their reputations, alumni networks (= potential clients), etc.

Small consulting firms usually build their reputations in local business and social circles, relying on repeat business and referrals.

Concretely, I'd suggest thinking about one or more of: (i) specific, rare domain expertise, (ii) finding one or a small handful of clients who can become comfortable hiring you repeatedly, (iii) research services that can be delivered in the form of data or code such that someone less expert than you can verify the quality.

What would an example project be? Be a little more specific on the services you would be providing.

I don't see why there shouldn't be a market for such a thing, but you need to be quite clear about what you can provide (your idea is a little fuzzy at the moment - possibly because you can offer a variety of things, but clarity and specificity are better for marketing)

True, I'm still working on sharpening the idea. Here's a specific example task, and at the upper end of the size/price spectrum:

Let's say you want to write a Twitter search engine and adopt existing FLOSS software as far as possible (my position a few years back). You could ask us to find, research, evaluate, possibly test components and fit them together according to your requirements. You receive an overview of existing projects, ranked by criteria such as maturity and completeness, all caveats for your project and a draft architecture using the best parts (crawler, indexer, front-end). This you can then hand to the implementation crew to get them started in the right direction.

Or you just want us to find out if anyone knows how to reliably capture all tweets from the system and try and get the information. I knew that back then, but finding it out took a mixture of ingenuity and perseverance - hacker skills :)

Here are a few rough ideas for a more specific service offering:

- software intelligence service: as above, plus similar tasks

- the scientific computing service: fixing bugs, improving display of results/export functions, writing a module, all the way to creating an entire solution (in Matlab, Mathematica, et. al.) In some sense this is an low-end IT consultancy with experience in scientific computing.

- "math-backup" service (for non-mathematicians in science and technology): have us search for a result, apply it to your domain/conventions, check for proper rigour in your work, find "bugs" in reasoning etc.

As RAs do a range of tasks in their positions, so could a VRA.

Sounds like a decent idea to me. Virtual research assistants work in a lot of areas, and I don't see why this should be an exception. I could imagine myself finding your services useful in the future. Some thoughts:

(1) If you're doing a lot of smallish jobs for people, the setting up of each job could add to quite a lot of your work time.

(2) (1) suggests you'll need to bill high per-hour to cover for the time you'll spend setting up, or try to do fewer, longer jobs (or ongoing series of jobs with the same clients). Another advantage of doing more work for the same person is that you will spend less time getting up to speed in a new area.

(3) This leads me to think the best way for you to structure it would be to take on a small number of clients, each of whom is likely to give you work fairly regularly. Not quite one-off, not quite employee. Freelance intelligence-gathering etc.

(4) Think about who exactly will be your clients. Academics, probably not. They have full time research assistants, if for no other reason than that's how departmental finances work most of the time. Large companies, probably not - they probably have interns and other such cheap sources of work-time. I think the one-man/small startup scene might be your market. Think of Patrick McKenzie (Bingo Card Creator - patio11 on HN) - he developed a relationship with a woman somewhere to write his bingo cards, simply because it's a better use of his time to pay her to do it. And he can trust her to do it consistently well - so it's like I said, not an employee, but not one-off work either (i.e. he will go to her, not look around for anyone).

(5) Perhaps you could frame your services in terms of competitor analysis ("Write me a report on all companies working on group microblogging"), which can range from business details to technical implementations, and so on; and also, as you suggest, possible relevant architectures for a project or feature a client may be interested in.

It might be hard to pull it off, but I think it could be useful to some people / teams. My somewhat HN-intoxicated suggestion would be to write up some example reports, whip up a website, and come back and offer your services. You might be worth the money to some very busy small teams.

It won't work. Do the calculation on the SIZE of your market, then do the calculation on how you will REACH your market.

You'll discover that there is a problem.

Care to elaborate? I'm not sure of the size, but there are 100.000+ related monthly searches on Google (under $.20 per click), which is how I hope to reach customers.

Searches for What?

I guess that depends on the specialisation (see other thread in this post). Maybe something like "software decision help", "scientific research service", "professional math help", "software evaluation", "Matlab debugging"...

It's a pretty bad and unscalable idea. Don't follow it. It won't work. 100.000 searches is nothing, start breaking it down to how many are willing to pay, and you see how small the market is. Unless you have an extra-ordinary passion for this area, find something more immediately lucrative.