As a study in speed vs skill vs
reliability, I'd like to prescribe
sylvie's
cute jump (windows only)
Consider the idea of uncertainty/unpredictability in the abstract.
If in a "simple" platformer focused on speed it feels like you're really just doing 'the right path perfectly', the only alternative is a mechanics that involves being hard to predict in some way.
Imagine a level with a hundred branches. If one is obviously the fastest, it is the one that someone concerned with speed will always take. If they all seem to take
approximately the same time and each one can be mastered independently in order to go a little bit faster, suddenly you have a game that will take much longer to speedrun.
Take this principle with you into a game with one "main route": how can a hundred equally-potentially-viable routes coexist in the same space? Maybe you have a boost that you can use once in a level; design four different places where a boost would really help with your speed, and now you have a choice, something to figure out: which of these places will save the
the most time? Maybe there are four different characters with wildly different movesets; if you design each level so that every character has a pretty viable way to speedrun through it, now you have another choice between 4 options. 4x4=16 routes.
With the examples you gave, think about how these present ambiguous choices to study as a runner:
your running speed increases if you consecutively jump on enemies' heads.
As your speed increases, you'll probably pass by certain enemies. So if I have four enemies in a row (ABCD), maybe they're tightly-packed enough that I can only really jump on one or two of their heads. Do I jump A->C or B->D? (Maybe it's also viable to slow myself down to hit A->B->C->D because of the long-term gains?)
in mario kart you can do tricks right after jumping off a ramp to increase speed.
This doesn't seem to leave much room for variance. It's just a straight skill test: do this, or you're doing it wrong.
general short-cuts.
Maybe a shortcut is harder than the regular path; is it worth it to risk my run on something I can only execute 50% of the time? 70%? 90%? (see cute jump)
Maybe there's a resource you get on the main route that you miss out on by taking a faster shortcut! Do I want to get an additional boost, or take this alternate faster path?
Maybe a shortcut is actually not even that much faster, and it's close enough that you're not sure which one will be faster once you've cut it down through practice?
something like collecting coins, and after a certain amount you can activate a boost for a period of time.
Put coins somewhere out of the way: maybe it's more difficult or more time-consuming to get those 3 coins. Is the cost of getting them now worth the boost you'll get once you have 10 of them? That's difficult math, but you have to design it so that it is actually difficult math!
You can stack everything on top of everything else. Maybe there are 3 boost coins on a shortcut that's much faster than the regular path but it consumes a boost to get into that shortcut in the first place. Remember that the goal is always to have as many viable (so: similar/same in terms of time) paths as possible! And also remember that this is an impossible goal, because there is always going to be an objectively fastest route. So you get as close as you can, waging a war against human ingenuity.