Thanks for the response. Which rule are you referring to here just so I'm sure we're on the same page?
Expertise. Many people seem to like it because it allows automatic success. Yet the rule itself doesn't mention automatic success or state it as the purpose of the rule; it's just a natural consequence of how the math works out, as early as mid-levels.
A couple of follow up questions if I may to see if I'm looking at it from your perspective. Do you assign every task the players describe their characters as doing a DC and ability check?
No, lots of things are automatic, if the answer is obvious to me.
Also, lots of things are opposed rolls versus NPCs, with advantage/disadvantage on either side for circumstances.
I often use NPC stats to generate a DC. For example, Deception against someone is often against their passive Wisdom (Insight), with advantage/disadvantage for circumstances of the lie. Sometimes I'll bump the DC up or down 5 for advantage/disadvantage on the NPC's passive. I find a framework like this is much easier for me to use in setting a DC than just pulling something out of my butt. If I do just pull a DC out of my butt, it's usually 10 or 15, sometimes 20 for a thing where I feel like the PC should probably fail.
Also, do you believe it's more interesting for the PCs to fail than it is for them to succeed?
No, but I think for a check to be interesting, both options need to be on the table. Otherwise we should skip the check part.
Combine this with my previous answer: if something is automatic,
I want to decide that, and the players usually go along. (Even if it's auto-fail, because I usually tell them beforehand. "You can ask the guard to step aside, but it won't work, because he'd be insane to do that.") If I start thinking about DCs, it's because the task isn't automatic, by definition.
So I don't like it when there's a modifier and a DC or opposed roll and then, because of Expertise, it becomes automatic after doing all that thought work --
it feels like Expertise is overriding DM authority.
I love "bounded accuracy" as both a player and a DM because it means that when there is a check, everybody has at least a small chance of success and at least a small chance of failure.
Do players ask to make ability checks or choose to make them on their own in your games?
I'm not quite sure what you're asking. Most of the times players ask to do stuff and if I don't have an obvious answer I ask them to make a check.
Here's another perspective, relating to the "DM authority" bit. I often view RPG play as a negotiation between the players and the DM. If the player says "I want to distract the chimera with a bit of food and then sneak past him!" it is almost always the case that the player WANTS to succeed, that is, they think they should succeed, or at least have a chance. Who am I to deny them that? I want to be a "Yes, but..." DM. So, I let the dice decide. I go to the dice when the players and DM don't agree that one outcome or the other "should" happen. The players get to influence the outcome in how they build their PCs, and by taking actions well-suited to their builds, and by spending Inspiration. I get to influence the outcome by picking the DC and advantage/disadvantage. Expertise makes my part of this process much harder because I can no longer set the DC without inadvertently allowing for auto-success or worse, auto-failure. In general I think the "auto" outcomes should arise naturally out of conversations with my players, or from discrete spell and ability affects that explicitly specify it.