I have a dexterity-based fighter-rogue using leather armour. Can I lose my dexterity bonus if I am hemmed in by enemies (cannot move to get out of the way) or surprised?

Asking this because Neverwinter Nights and other DnD 3.5 games have a skill called tumble that adds to your base AC every 5 levels. Both rogues and bards get this skill (and monks). When you get to a high enough level it enables you to be virtually untouchable if you keep moving all the time.

Does IWD2 have the same sort of mechanic "under the hood"? Or do you always keep your dexterity bonus?

Thanks - your explanation is consistent with the threads you can find along the lines of "(heavy) armour in IWD2 sucks". That's why high Dex and light armour (together with buffs) can beat heavy armour even when the 3.x rules would make you lose your Dex bonus.

So on that issue it is more like 2nd Ed rules.

This isn't the reason why heavy armor is bad in IWD2. The issue is two-fold:

First, IWD2 uses a method of distributing stat points that makes it really easy to have 18 dexterity in addition to 18 strength. In actual 3rd edition character generation rules, it's rare for a character to have even a single 18 much less two, so no strength-based melee characters would have a dex score that high. The 3rd edition armor types were balanced on the presumption that a strength-based warrior probably only has about 12 dexterity, and the max dex bonus was intended to have little to no impact on those characters.

The second issue is that IWD2 includes a number of medium and light magical armors that give excellent AC while also allowing you to benefit from a high dex score, but heavy armors were all capped at around 12 dex, so there was no heavy armor that worked for a character with 16-18 dex. This meant that pretty much everyone ended up wearing medium armor instead.

This is simply a consequence of IWD2 being in production while the 3rd edition rules weren't even officially published, so the game designers were working with a brand new game system they had no experience.

Baldur's Gate 1&2 also has PCs who are stronger than they should be, and that was released 9 years after 2nd edition was published. I think the IE games do it intentionally to attract people who don't play P&P.

I would disagree; the (AD&D) IE games allowing you to more or less just pick your own stats was a canny understanding of the underlying flaw in character generation in AD&D. The AD&D ruleset was originally built around the presumption of a Gygaxian adventure, in which adventurers begin as ordinary people, go on adventures that will result in most of them dying horrible deaths, and a few live long enough to retire as noteworthy heroes. If you look at AD&D character generation through the lens of a Gygaxian adventure it makes sense; if you roll 3d6 you'll find you'll get no meaningful bonus or penalty the vast majority of the time on the ability score tables, and on the rare occasion you roll high the bonus you receive isn't high enough to cause severe game imbalance. Baldur's Gate is not a Gygaxian adventure (which had fallen out of favor completely by the 90's anyways); it starts with the presumption that you're a special prodigy destined for greatness. The decision to just let you pick your own stats makes sense in that context; if the game presumes you're a prodigy, then you'd need to have some 17's and 18's, so at that point you may as well go for broke and allow the player to just pick what they want.

3rd edition changed things up. It did away with the presumption that most adventurers were average, and created ability score tables that gave a better spread of possibilities. Now bonuses from high ability scores kicked at numbers as low as 12 and scaled up at a regular pace. You no longer needed to give players any ability score they wanted to make them prodigies; a 16 makes you fantastic with significantly better bonuses than an AD&D character would have with an 18. This meant that letting players have 18's across the board caused significant balance issues in a way that it didn't in AD&D, and wasn't necessary because you didn't need an 18 to feel special anymore. In this way, using this method of ability score generation represents a misunderstanding of the game system.

That is a perspicacious observation. :)

I haven't played AD&D for over thirty years and what you identify certainly does ring a bell. (I had evidently elided all the AD&D versions, with the later releases obliterating the earlier gameplay, but I recall vividly that bonuses were fewer and less frequent.)

I suppose this could also be construed as a natural evolution of the (traditional Gygaxian) AD&D character to ameliorate the inherent weakness of the character system, like the omission of demi-human high-level character classes. With the systematization of bonuses, regulated (positively) from twelve and no longer limited (for demi-/humans) to a maximum of eighteen, one could extend the system to incorporate almost limitless variety.

Thus the (JRPG-inspired) boundless bonus additions, stacked for multiple items and feats, but keeping the icosagonal engine, to keep critical failure at 5%. And the level-derived bonus, where experience grants an extra rank in any of the six abilities; now eighteen isn't exceptional, it's almost obligatory (especially for a magic user, divine or arcane).

It seems that a lot of these "chosen one" narratives are some sort of committee-think to explain the whole process of gaining abilities with levels (and especially high-level proficiencies) in games like Planescape and Baldur's Gate.

I agree, though, that we have lost something with the passing of Gygaxian ordinariness. Maybe it will come back into fashion?

edit: word-processing

In all fairness my experience with AD&D is academic, having read up on it out of curiosity. I was too young to have played it in its heyday, but I do find it very interesting to see the evolution of tabletop roleplaying games. It's absolutely fascinating to read an in-depth description of an actual play session from AD&D or OD&D and see all the elements that are still in the game today, and the others that have been discarded or modified beyond recognition. While AD&D did put out additional rules supplements to cover things that the standard rules did not, they did so in a very haphazard and inconsistent way, so over time the expanded AD&D ruleset became an incomprehensible mesh of overlapping or outright conflicting rules. It's said that by the time of AD&D's dying years in the late-90's that every DM had a binder full of their own rules clarifications and addendums, and every DM had their own distinct version of AD&D that would be almost unrecognizable if you'd learned the rules from a different DM. You definitely can run a Gygaxian campaign in 3rd edition. You just need to keep ability scores low and retire before reaching the higher levels. The rules aren't specifically designed for it, but they're a lot more flexible than earlier D&D rules so you can modifying it to suit your needs.

The bigger issue is that Gygaxian adventures simply aren't that popular. The high mortality inherent in Gygaxian adventures means that players can't get too attached to their characters, and it's hard to build overarching plots when the key actors die at random times. Modern tabletop gaming tropes have emerged for a reason; you need a certain amount of "plot armor" in the main cast to actually have a stable long-term narrative.

With that said, the "chosen one" trope isn't used consistently. Sometimes it's embraced, sometimes it's discarded, sometimes it's just alluded to. For instance, I'm currently running a pre-made adventure called "War for the Crown", a story about a succession crisis and civil war, where the player characters serve as operatives working for one of the candidates for the throne. The player characters aren't chosen ones in any sense, they just happen to be talented people living in interesting times who have a chance to alter the course of their country's future.