This post builds on my Don't Trust Speedrunners! and Mechanical Contrast posts, so check those out for more context! If you want somewhat of a positive defense of the value of skips, check my Organic Exploration post.
A lot of players tend to love skips & movement tech, especially in platformers. And on the surface, it seems like they're unambiguously good - they allow for extra skill expression, they widen the skill gap, they make players feel clever, they have nice feedback to them, they seemingly add a more robust routing element, etc. It's only when skips completely override the rest of the game's routing too heavily do players start being concerned, creating no skip categories and stuff like that.
In my view though, in any game with fairly deep, continuous movement, skips are mostly detrimental. If you're making a game like that, I believe you should put a lot of effort into making sure that players have to stick to the "track", and engage with that track as much as possible. You need a really clear framework for what makes a good skip, and what makes a bad skip, because not all of them are equal. Or at the very least, you need to recognize that an emphasis on skips creates a different style of speed-game, similar to linear vs exponential scoring in shmups, rather than creating a game with a higher skill ceiling.
Consider the left side of this post's picture. Imagine a simple Marble-style game where rolling downhill builds speed, rolling uphill decreases it, and going as straight as possible preserves any speed you've built. On the left side, you have a more straightforward route some players might default to. Every single slope can be seen almost like a node (though it's actually more complex than a node!) - a decision point where players figure out which direction to turn in. Depending on where they turn, they might save or lose time. The time gains/saves are extremely small, usually being measured in milliseconds. These optimizations only become truly exciting & intense at higher levels of play, meaning they are pretty invisible (or irrelevant) to newer, and often even intermediate players.
Now consider the right side of this post's picture. You have a similar kind of Marble game, but players have found an exploit. They learned that by doing some dark magic trickery at the corner of the slope, they can launch the ball forward great distances. The exact distance depends on their execution of the trick, so there's a lot of ways to fail or succeed at this trick - it's deep. Successfully performing this trick will save potentially dozens of seconds, while faliling will make you drop off the level.
While this trick might provide an initial burst of excitement, and might create an instant easily recognizable skill gap between beginners & experts - it's an illusion. Instead of the game being broken up into many different "nodes" where micro-optimizations are possible, the game is broken up into a single node or a small set of nodes where micro-optimizations in inputs correspond to macro-level time gains. All skips do is make the difference between bad & good performance more clear. They don't actually meaningfully add skill expression into the games, and often outright remove it.
You can probably see how the depth/challenge level of the skip itself is pretty much irrelevant - it can be extremely deep & difficult, or extremely shallow & easy. You have to evaluate it in relation to what exactly is being skipped, and the potential depth/challenge of that content at high levels of optimization.
You can think of this in terms of Opportunity Cost - a skip's quality depends on whether or not the player is skipping parts that create more room for skill expression. It's possible to separate skips into two categories - High & Low Opportunity Cost Skips. The less deep game's general movement is (and the less your level design capitalizes on it), the lower opportunity cost there is to the types of skips outlined in my example. That's where skips adding more depth & skill to games makes sense. Otherwise? Forget about it.
