[Oblarg]

As a drive mentor, this is pretty much the single most important design constraint I deal with in FRC. I don't care how impressive-looking and feature-packed your drive is; if at any point it fails during a match, it has cost you more than the added features could have possibly given you over a simpler design.

Keep it simple, keep it durable, keep it serviceable. You cannot break any of those rules, ever, if you want your drive to do its job. If you've got a choice between overbuilding and underbuilding, always choose the former. It is far better to have to cut features due to weight constraints than to have your robot break down.

Keep in mind that "simple, durable, and serviceable" does not mean "unambitious" or "trivial." I've seen many ambitious, nontrivial drives executed in an elegant, robust manner. Most FRC drive concepts can be implemented in an extremely reliable way if you execute them properly (though a few, such as swerve, may require somewhat prohibitive team ability and investment of resources); it is, as always, a matter of details. But, as a rule, if you ever find yourself doing something which looks at all like sacrificing reliability for added features, you are doing it wrong.

An afterthought: If you are a team with durability issues, and you tend to make lots of parts out of 80/20, the two are very likely related. 80/20 is a fantastic prototyping material. It is not a material for finished robots. I learned this the hard way during my time on 449's drive team; no amount of tightening, loctite, or lock washers will keep things in t-slots from eventually coming loose. Fix your dimensions and attach things with through-bolting or pop rivets.