There are a bunch of ways:
* The success or failure of fights often hinged on a single die roll for powerful abilities. Besides metagaming hard and soft counters after a reload (which I'll get to, below), these were elements where the player's choices did not have a ton of impact; their success or failure mostly depended on the outcome of a die roll. In some cases, there's really a tiny set of hard counters (e.g. Protection from Petrification for use against basilisks). Most other tactics just shifted odds and asked the player to hope for the best and reload if/when the worst came true. Reloading is a part of these games, but I don't think anyone wants it to be a core mechanic for success.
* There are many bad ways to build characters in virtually all of the IE games. Leveling was a little easier pre-3E, but you could make an absolute garbage character in 2nd Ed. very easily. Players should learn to play to the strengths of their characters, but with many builds, there was no strength to that character -- just a lot of suck. 6 characters * bad stat arrays = a slow but steady descent into a non-viable party near the late game. This was mitigated somewhat in the BG games since they were balanced around the companions, but it was a huge problem in the IWD games.
* Pre-buffing alters the difficulty of fights enormously. About halfway through IWD's development, a QA tester (who went on to become a pretty well-respected developer) came up to Black Isle and was furious at the difficulty of a fight in Lower Dorn's Deep. He had been trying to legitimately get through it for 2 hours and hadn't succeeded. Kihan Pak and I loaded it up and beat it on the first try. He asked to see what we were doing. Naturally, we were pre-buffing for 5-6 rounds before we even went into the fight. Because there was no opportunity cost to using buffs, this was "the way" to get through fights, but it was tedious -- and for people who were not D&D veterans, it was not something they ever thought to do, which resulted in a full roadblock (see also: Burial Isle misery, which was also pretty easy for me and Kihan).
* Hit points make the world go 'round. There are specific party and resource builds you need to maintain your hit points over several fights. If you don't use those party builds, you suffer enormously or have to backtrack and rest very frequently. This is one of the major reasons why we have a split Stamina/Health system for short term/long term damage (and why 4E uses healing surges). In PoE, getting knocked out takes you out of the fight, but when the fight ends, you're still in the war. In IWD, if you got super-slammed and weren't ready to devote your precious healing resources on getting that dude back into shape, you had to pack up and head back to a safe zone -- or rest on the spot and reload if you got an encounter, which isn't much better. Along the same lines, almost all character resources that were limited were per-rest, so if you used any of them it was a big deal. Per-rest resources are a big deal in PoE, but every class also has per-encounter resources as well.
* Many fights could end in Pyrrhic victories due to level draining, petrification, or character gibbing. There's a fine line here between an interesting tactical/strategic element (i.e. how will I deal with this affliction in upcoming fights) and something that 99% of will simply reload after experiencing. Some of these things can be toggled by player difficulty settings, but other elements can be redesigned to still be interesting without being obnoxious. And again, many of these things that happen (especially with long/permanent durations) rely on raw luck or the use of hard counters that the player needs to reload and metagame to prepare for. A Dire Charm with a long duration is (if you lack a hard counter) essentially an immediate KO for that party member and bolstering of the enemy ranks by a character of equal strength -- for the rest of the combat.
* Stand-alone random rolls are pointless outside of an Ironman-style mode. Random resting encounters, rolls to learn a spell, rolls to pick a lock, etc. The player is better served by having those things be thresholds (or non-existent) and giving them tools to increase their ability to meet those thresholds. Failure to make a stand-alone random roll is not a failure on the part of the player; they just got a bad roll. You can get bad rolls in combat, too, but those are part of a big shifting soup of randomized results hat happen over time.
Now, these are all things that clearly a ton of people adapted to and worked around. But it should be asked: was adapting to them interesting and enjoyable or just something that you did so you could enjoy the other parts of the game? If the latter, we should really try to find ways to not repeat those things in PoE.
In what ways were any of the IE games' challenge "oppressive or frustrating" ...unless he's talking about pathfinding.
