That's like the town bell ability in Age of Empires III. Basically, there's a button at the town hall you can click if your base is being attacked, to make all your villagers run and garrison inside the town hall or a watchtower (whatever's closest). When you unclick the button, they all run out back to whatever task they were performing previously (it's not perfect, but definitely saves on re-tasking time).
While Det and Brent are mostly right, I think that in conventional Starcraft-style RTS, some automation is appropriate sometimes. But in that case the criteria are a little different.
Basically I think it's worth looking at (note that I'm saying looking at, not implementing - you would still have to judge implementation on a case-by-case basis) automating any tasks that are mind-numbingly tedious.
@ Brent: I'm on record elsewhere as being a member of the camp that says luck is an integral part of games. Like filthy socialism to pure Godly capitalism, the exact ratio of luck to skill in a given game will always be a measure of vicious debate. But it has it's place.
Not just so a designer can pander to the CRITNOOB GLORY crowd, but also because a) there are skilled players who love battling impossible odds and winning. Not just made-up 4v1 handicap matches, but a REAL impossible situation no one could foresee, and b) In real life, shit happens. Or CRAZY WHAT THE HELL I DON'T EVEN happens. Some people genuinely like games that reflect that fact.
Not saying every game should have a luck component, or that everybody enjoys 'a' or 'b' but the more variables a game has, the more I lean toward having some random element involved.
The thing is Mongrel, when you are playing in a game that does not have perfect information, randomness doesn't have much of a point. You don't know what your opponent is doing, so you have to scout to find out. There is considerable depth right there. What would be the point of having that depth if you might not benefit from your intel with random damage, or whatever?
In essence, having a game where you do not have all the information you need to make informed decisions is the perfect kind of 'randomness' a competitive game can have and still be good.
Basically, if you're aiming to make a competitive game, you want as little randomness as possible (or at the very most, very well considered randomness in small doses), preferably none.
If you're just making a game for general consumption where you don't care if it lasts for years and years (and years and years), then go ahead.
TF2's sorta a bad example in this regard as the community is divided over modding out crits or not; dividing the community isn't a good thing.
-
Also, you shouldn't be spamming bunkers to the extent where you would need to have that sort of automation. It encourages players to turtle more than they should... and usually you don't want just whatever ol' unit to get into that bunker, you want to pick specific units due to low health, or damage output you want to protect. It's not like the AoE3 or War3 examples where there just this one type of unit that is useless otherwise and of course you'd want to put them in the structure where they can fight. It's a completely different situation!
Consider the Peon. Nearly 100% of the time you would want to get him into that protected structure where he is safe AND deals more damage. There isn't much of a game there.
Now consider the Bunker. You need to decide what ground units go into it, and this can be an important decision. It's not a no-brainer like with the Peon. There is a game here. You
could add the automation anyway and deal whatever units get into the bunker first, but I see that as a crutch for bad play that I'd rather not encourage. Be glad you have production queues!
And auto burrowing is bad, out right. Sorry! It would require a total revamp of the mechanic which isn't worth it as it's clearly proven itself to be well balanced.