I think it's interesting to relate this thread to Fred Wilson's "Why Early Stage Venture Investments Fail":

"Of the 26 companies that I consider realized or effectively realized in my personal track record, 17 of them made complete transformations or partial transformations of their businesses between the time we invested and the time we sold."

With this in mind, I think the most important factor is tractability. Can you build it quickly to test your theory? If your theory is off-base, can you quickly shift? I think Fred is a fairly discerning and successful investor... It's interesting to note that 2/3 of the successful teams that he has invested in have been WRONG about their initial product direction.

I'd prefer to evaluate my product using the method of the other EW (me):

if (customers.want) {return success} else {return regroup} ;

A better question: will it blend?

Tarsnap (online backups for the truly paranoid):

Tractability: Medium.

Obviousness: High.

Deepness: Medium.

Wideness: Medium/High.

Discoverability: Low.

Monetizability: Very high.

GameClay, online casual game creation.

Tractability: Low

Obviousness: Very high

Deepness: Medium

Wideness: High

Discoverability: Very high

Monetizability: Low

I think you're selling yourself far too short on monetizability. You could have a model where ad-supported games are free to make (i.e. YOUR ads), or where they could be ad-free (or contain their own ads) for paid subscribers...

Monetizability: Very High

after 10 long minutes:

tractability.Med obviousness.Med deepness.High wideness.VeryHigh discoverability.VeryLow monetize.VeryHigh