Sounds like a punk band...
Anywho, occasionally the Cathode Tan inbox gets a question. Usually it's things like "Would you like to make more money?" (A: Only if you're willing to give me all of yours, no questions asked) or "Are you really my daddy?" (A: No, I just said that because I stole your flag). But sometimes, it's a good question:
This is actually an excellent question, especially for these days. I was going to just send the man a quick answer and then go into a longer diatrabe here ... but the quick answer turned into a longer diatrabe so I'm merely reposting now.
The biggest pro of modding is that you get to play with an extremely powerful engine, free of charge. For ease of use combined with power - the Unreal engine is simply the bomb. There's still a vibrant community and Epic has been heads and shoulders better with supporting modders than any other company on the planet, bar none.
Small mods, however, are generally overlooked these days. Unless they either a) fix something a lot of players don't like (such as the Doom 3 duct tape mod for using a flashlight with a weapon) or provide some early toys for people to play with once they've finished the game (like Garry's Mod, or to an extent - my original Xpaks for UT2003) - most gamers may try a smaller mod a couple times but that's about it.
Even larger mods, like Red Orchestra, are having a hard time finding an audience.
Course, the advantage is that there's an installed user base. One of the reasons Counter-Strike got popular was because virtually everyone already had Half-Life. So while it might be near impossible these days to make a sustained audience - you're pretty much guaranteed that quite a few will probably at least download it.
Now I'm pretty new to indie dev - but it's more or less the opposite. You get a much less powerful engine at cost (though there are free or cheap alternatives to that) ... or you end up spending a lot of time coding the engine itself. Community support will be much more scattered (but present, and actually in some ways I'm finding indie devs even less cutthroat than some mod teams).
As for downloads, I haven't really had a chance to prove it - but basically you're up against actual market pressure. So you've got to have something to "sell" in order to get any sizeable amount of downloads. So if you make Pong for the PC, I wouldn't expect many hits. Course, if you made Pong for some cell phone or PDA that didn't have any games on it...
But there is one big advantage to indie dev. You own what you do. Even if it never sells, you can point to it and say "that's my game".
If you're looking to blue box something - it gets a little complicated. The first obvious question is ... what engine most closely suits your idea.
If it's like an RTS, I'd probably go look at the Torque engine ... most RTS games don't provide mod tools for serious TC development (unless you can work within the restrictions of the original game). I'm not intimately familiar with the Torque RTS kit, but it seems more flexible than say ... Freedom Force (which is really only suited for making new FF campaigns).
If it's a shooter or other 3D action type idea, then Unreal is probably an easy choice. While there are FPS alternatives in the indie world ... their only advantage is the license. Course, if this is something you intend to shop around to publishers - the license might matter. But if you're just looking to "prove" some ideas in a 3D gamespace ... Unreal is the way to go.
So the really short answer is - 2D stuff? Go indie. I would highly recommend Torque 2D. 3D stuff? Go Unreal. If it's 3D stuff you intend to sell, go Unreal but don't go too far with it. Just far enough to gather a small team to port to an affordable engine (Torque, Reality, etc).
If it's nothing you intend to sell or profit from, and you're just looking to prove that something works or that nobody tried X in an FPS to see if Y works, then god bless you, head over to the Unreal community and get to work.