Wicc
Auto-translated
Dima Syurmey's idea of limiting the maximum battle time is interesting. You need to defeat a horde of golems, and it definitely won't be possible to do it in a minute or a minute and a half. So, the limit would be, say, 4 minutes, even if the battle lasts 10 minutes. Otherwise, one such battle will ruin the entire chess timer.
The average inconsistency of the generator significantly exceeds the inconsistencies in battle length.
I remember, on a fixed MvM map, I developed a strategy to take the dungeon on day 111 with a bunch of dwarves near the castle. This was on TE, without morale, with a timer of 1800+480 and a random starting hero. There wasn't much money on that map, but I wanted to do a lot, so the idea was to get a lot of money for the initial development at the cost of bringing the timer to zero if the dwarves are tough (taken with a group of harpies). Therefore, in addition to the technique of executing the battle, I also had to prepare a scenario for the entire first week, making screenshots with notes on where to place the heroes on which day, etc. This was to save at least some time for the second week. The strategy turned out to be quite strong, but extreme, requiring quick reflexes and good theoretical preparation.
With a sodotimer or a limit on the battle duration, this would be a must-use strategy, and the first day would be played for a very, very long time.
Added 5 minutes later
By the way, regarding the bonus in the late game, I remember that in WT there was a proposal to make the timer three-tiered - starting time, weekly bonus, daily bonus.
Using the example of the current timer, this could look, for example, like a double bonus on the first day of each week. That is, if the timer is 16+8+1, then on day 121 (131, 141, etc.), 16 minutes are added instead of 8. This will not significantly affect the overall duration of the game, it will not eliminate the need to save time, but it will slightly reduce the hassle in the late game.