This has never registered as a problem to me because so many tasks are already automatic passes or fails anyway. I never ask for a roll unless the task has a reasonable chance of success, a reasonable chance of failure, and a consequence for failure. If your rogue didn’t have reliable talent, but still had his 20 Dexterity, +4 Proficiency Bonus, and Expertise with Thieves’ Tools, what’s stopping him from picking every lock with a DC 23 or below anyway? Do you not allow players to take 10 when they have no time pressure? Heck, even 33 or below, unless there’s something preventing him from trying over and over again until he succeeds, he’ll eventually roll what ever target number he needs. During uptime there are usually complicating factors that prevent characters from taking as much time as they need on tasks, but if you’re handwaving that away on the basis of “what if he wants to rob a town blind during downtime?” then what was stopping him from doing that before? And anyway, at level 12, this rogue should be a grey mouser type legendary thief. He should be able to rob entire towns in his downtime, that’s the fantasy people play rouges for.
At the end of the day, Reliable Talent just allows high level rogues to take 10 even under pressure. That’s nice, but it’s not game-Breaking by any means.
I don't allow taking 10, no. It's not in the rules. If you're going to allow unlimited retries, don't roll, just pick a time window it takes and move on.
If I ask for a roll, and you fail it, there's a consequence. If I ask for a roll on a lock and you fail it, you've either damaged the lock, increasing the DC or preventing it's picking at all, or broken your tool, or some other appropriate failure. I no longer run in a way that allows 'you failed, try again.' That's boring.
My issue with reliable talent is that it badly interacts with expertise and that it's a sudden, massive improvement in skill execution. A rogue with a +13 check goes from failing a DC 23 check 45% of the time to never failing it in one jump. I'd prefer a shallower improvement curve. Also, it completely defeats the advantage/disadvantage paradigm, which I find poor design. That same character trying a DC 23 check with disadvantage goes from a 70% chance of failure to a 0% chance of failure at the same task with reliable talent. Don't like
Secondly to the above, it's a kinda hidden boost -- it doesn't appear on a read over of the class to be as strong as it is, and so new DMs that are crossing the 12th level line for the first time engage in a sudden bit of shock when it shows up and obviates entire classes of challenges that were, just a moment ago, still part of a fun adventure.
Can I build around it? Sure, if I distort a bit of my fiction, and, hey, what about D&D doesn't distort the fiction. But, if you're not planning on it from the get go, it's a pretty big and sudden distortion. Locks go from something that present a minor challenge to something that presents no challenge, suddenly. Dealing with that along with learning to run higher level games at the same time is a pretty big hit in the cognitive balance area. I dealt with it the first time as an experienced DM with a bit of trouble, but got over it quickly on my own and adjusted my game. In my current game, I've instituted the above houserule about disadvantage, but otherwise left it alone. The player, mostly, chooses to engage disadvantage or not, so it's on them, not me, and that seems fair.
As for a easy fix for locks and the rogue cleaning out small towns -- most people can't afford locks, but they can afford to bar their doors. And you cannot pick a bar on a door. Depending on construction, and where the bar is latched or not, it may still be easy to undo, but you can wildly vary that effort with better construction or a latch on the bar rather than trying to justify wildly out of place and expensive locks. Also, it's difficult to remove a bar quietly.