I think game design can happily accommodate true teamwork along with managing the potential risks of random coop. One idea that immediately comes to mind is a slider for 'cooperativity'. Each player has a range of skills or equipment that serve specific purposes, such as lock picks, scanners, jammers, smoke grenades, etc. The cooperativity slider would determine how much variety a player could yield themselves. With it all the way on 'solo', a player could equip items and skills to solve all issues. With it all the way on coop, each player is limited to a couple of options, meaning a dependence on the others. Half way, that dependence would be reduced but still there.
The fact games aren't managing a good balance that supports true coop is just because devs aren't really trying IMO. In gaming there are those who want to solo, those who want coop as in random games where you never know the capability or interest of the rest of your party members, and those who want to play with a specific group. The first two options are covered by creating a solo game and then allowing other players to join with the same mechanics. The latter bunch are really not catered for one jot, but there are definitely systems that could be employed if developers would just investigate them.
That's an interesting idea. But it makes the skilltree's, something which games rely on to give them longevity, somewhat obsolete. Mapping a lot of skills and equipment to single control scheme for the 'solo' player may be difficult. And I could see the 'co-op' players all wanting the same equipment and skills, and all not wanting some others. But perhaps these things can be designed around. I agree that developers aren't really trying with co-op. It's usually just something that's added as an afterthought.
Fuse at least seems to be primarily about co-op. It's kind of crazy how little we know about it at this point when the game is only half a year away from release. They didn't even bother to show it at this year's E3, or at GamesCom.