I'm new to the forums and IA, but I have a few ideas that might work. First, I should start with the things I think should probably be added, in keeping with the spirit of the mod.
Since Sikret has expressed distaste for commonly used cheats, a cheat being defined as an exploit of the game's engine that a DM would not allow in PnP, I want to call attention to the "cast-and-attack" trick. This is illegal in PnP and makes little sense from the perspective of realism. Fighting and casting spells both require extensive time to perform under the pressure of combat. It also works to the distinct benefit of fighter dual- and multi-classes, at the relative expense of single-classes, though I've gotten the impression that single-classes in IA are more viable overall, at least compared to multi-classes. Casting spells and attacking should therefore be disabled in IA if possible, and if it is not possible, a simple reminder to the player in the readme or one of the in-game hints would call attention to the matter.
I realize that the trick is popular, and I myself am very fond of it, but it makes sense in the context of IA.
Second, I'd like to criticize PFMW as a tanking spell. Many creatures in IA appear to lack normal weapons and hence have few options when dealing with characters with PFMW. The only way to counteract it is Breach and RRoR if necessary, which not all enemies should probably have and which seems a tad monotonous. The spell effectively makes the caster the perfect tank against any enemy fighter type for four rounds, giving the caster a great deal of time to cast spells in safety. Preventing the caster from casting it more than once every few rounds (maybe 2 turns, so it could not be chained) would make it less overpowered. Absolute Immunity has a similar problem, though arguably it is less overpowered, being a 9th level spell.
Other spells also would be better balanced by weakening them. Reducing the strength of some resistance spells (Protection from Fire, Protection from Magical Energy) would make item or Vagrant resistances less redundant, and give the player a reason to memorize Protection from Fire rather than, for example, Flame Arrow. Lightning Bolt also seems rather overpowered, though few seem to use it. Many areas in IA have close quarters and it is easy to get lightning bolts to bounce correctly and hit an enemy four or five times: the player can just fire the bolt directly north or south. In the vanilla game, I've found Wands of Lightning and Lightning Bolt spells to be more than capable of dealing with almost any enemy, and they still enjoy an advantage over Flame Arrow in many situations in IA, even if they are not as strong. Mirror Image and Stoneskin also are considerably more valuable than most spells: there is little reason to memorize Blur or Teleport Field instead. Feeblemind is another example, effectively being a low-level Finger of Death that bypasses Death Ward. Making the duration temporary and/or tweaking its effects (IWD2 just removed spellcasting for 2 turns) would make Hold Monster a more viable alternative. Offering a higher- or lower-level Breach equivalent would make other 5th-level spells more worth memorizing (how often does anyone use Shadow Door or Domination?). Emotion's excessively high duration also gives it a great advantage over other disablers. There is little reason to use Hold Monster or even Chaos when one has access to Emotion.
At the same time, other spells receive nerfs that make them nearly useless for many important encounters in IA. With enemy MR and its save bonus and low duration, Glitterdust has few uses in IA. Generally, I think a broader variety of useful spells would be suited to a tactical mod.
I think we should consider re-introducing (Minor) Spell Deflection and Minor Spell Turning. Currently the player has few spell protections at his or her disposal. Creating some Transmutation spells to replace them would also work, and have the additional benefit of balancing mages, since transmuters miss out on so many excellent abjuration spells. Arguably, these spells give the player an unacceptably strong defense against enemy spells, requiring fewer spell slots than Improved Invisibility+SI, but the enemy would also be able to counter them with spells aside from RRoR, if we also included some lower-level debuffers that IA overwrites. Making Pierce Magic and even Secret Word or Spell Thrust more practical would balance mages versus sorcerors by making a broader variety of spells more valuable.
I think it is important that we make a broad variety of spells useful. A tactician would naturally want to be familiar with many different methods, and I understand that the tactician ideal is a big thing for IA players here. Extra options can help give more color to the game and make some battles less monotonous. I have found it unfortunate that the strategy for dealing with mages in IA usually boiled down to RRoR, True Seeing, Breach, and Chain Lightning. In IA, Wizard Slayers, Insect Plague, Silence 15' Radius, Miscast Magic, backstabs, or even simply using a Fireball to interrupt a spell, are all of very little value. Such strategies are just not worth pursuing in IA considering the strength of RRoR, True Seeing, Breach, and Chain Lightning, and I think it would be nice if those tactics were useful at least in some situations.
Other times, I've found spells to be useful only in very specific, unique circumstances. I was delighted to find that Animate Dead was useful before level 15 in dealing with the rats in the Lilarcor quest, but unfortunately, that spell had little use elsewhere, and so I never bothered to memorize it.
There are other spells that IA overwrites that could be reintroduced without damaging game balance or weakening difficulty. Otiluke's Resilient Sphere could be weakened by offering a saving throw bonus and lasting several rounds less than the vanilla version, which would give the player another tool to deal with enemies that was not overly powerful. In vanilla BG2, ORI was much like a Maze spell with a saving throw. Project Image could be modified to either prevent the image from casting spells or drain the caster's spells when the image did cast spells. This might be difficult to implement, however.
One might argue that reintroducing these spells would either require increasing the number of spells overall--preventing the sorceror from picking all of them--or require overwriting useful new IA spells. This could be avoided by replacing less-commonly used spells such as contagion, which few players (I assume) bother using anyway.
Moving on to IA-specific changes, I would call attention to IA kits' hefty advantages to other builds. The Auramaster has essentially every advantage over other druids. It suffers in melee power and HP, but then, druids aren't there for dealing melee damage anyway (none of them are very good at it), and the Auramaster's Mirror Image and extra castings of Wondrous Recall, Ironskin, and other buffs make it a much better tank than the other classes. Its disadvantages are much smaller than its advantages. A druid is there for spellcasting, and the Auramaster is simply better at spellcasting. Its access to RRoR even makes it a strong alternative to a mage--a significant leap for a divine spellcaster. There is therefore little reason, from a tactical perspective, to pick an Avenger or Shapeshifter over an Auramaster.
Much the same holds for the Vagrant and Riskbreaker, along with vanilla kits such as the Swashbuckler and Necromancer. Their advantages appear to more than outweigh their disadvantages relative to other kits, making other kits more or less pointless. There is a reason why IA players tend to use IA kits or classes which have access to IA-only items. If we are to even keep the other kits, it would make sense to strengthen them or replace them with more sensible alternatives to the Vagrant et al.
Much the same holds for IA-modified and IA-introduced items. It seems that much of IA character advancement is focused on, or even dependent on, the player collecting ingredients for Cromwell. They give the player something to go after, but they are so strong that it punishes players who want to stick with vanilla weapons. I realize that IA is very much a new game, but making IA items less overwhelmingly superior to vanilla items would open up some new opportunities for players, which fits the point of a tactical mod.
IA also lets SI block the thief's detect illusions ability. As I understand it, the rationale is twofold: first, the ability in vanilla BG2 is overpowered, as it gives a low-level the ability to dispel any mage's illusion spells, no matter how skilled that mage may be. Second, SI's blocking the ability is justified because it blocks divination "tools" and not simply spells. I have two objections. First, detect illusions can be nerfed without rendering it incapable of helping the party deal with mages using SI:Divination, and all mages in IA who use Improved Invisibility--the most troublesome illusion spell in IA, I think--always use it with SI:Divination, which makes the detect illusions ability mostly useless against important IA enemies. To weaken it without rendering it more or less useless, we could edit the 2da file for racial thief bonuses to give all races a -150 penalty to detect illusions. So, in order to get detect illusions to work well, the player would have to expend 10 thief levels, 250 points, on detect illusions, which I think would be an acceptable sacrifice. We also might give the detect illusions ability a certain chance to fail, even if the thief's skill is 100.
The ban on traps is also limiting. The objection, as I understand it, is that the enemy cannot set traps to harm the player, since the player already knows where all the traps are and can disable all of them, though I believe the dispel magic trap in the drow ambush is an exception. I would advocate instead that the enemy gain spell-like abilities that would mimic player traps. That way, the enemy could have thieves do something besides backstabbing, adding some variety, and it could justify giving the player another tool without giving the player a simultaneous and unreasonable advantage against all enemies. Traps can be fun, and they are one of the reasons thieves are actually useful in BG2. With many enemies in IA immune to backstabbing, the thief often has little utility in combat unless it is a Swashbuckler. Time Traps might still be better off being disabled, or at least weakened, for the sake of balance.
Adding some content for protagonists besides Vagrants, good fighters, Necromancers, Swashbucklers etc. would also balance the classes. I realize that this might require a lot of extra work, but it also might be helpful just to expand protagonist-specific bonuses to other kits: for example, letting non-Necromancer mages get access to their own version of an Amulet of Hades. Adding some content for certain NPCs would also make them more viable. While Edwin had essentially every advantage over Nalia in vanilla BG2, the trend is reversed and is equally unbalanced. Adding an upgrade for Edwin's amulet would help. Much the same holds for Minsc vs. Valygar; Valygar is by and large the superior character, given his IA bonuses. Given Minsc's popularity, I would actually have assumed that Minsc would enjoy the extra attention.
I have found the early game fairly frustrating, and I have gotten the impression that players have gotten alienated from IA when they find its battles too difficult. The fact that many battles that were easy in vanilla BG2 are incredibly hard in IA should only exacerbate that problem and deprive IA of the attention it deserves. I struggled through many encounters before realizing there was a recommended order of quests list. I ended up doing the quests in much the same order as other people have, but only after much time was spent. Including the recommended quest order list in the readme instead of the forums would help new players adjust. Goodness knows the work of Sikret and IA testers isn't appreciated much if few people play the mod all the way through. Adjusting the difficulty would help bring in a bigger audience. I have gotten the impression that most people find that the late game is actually noticeably easier than the early game, and that some party builds just aren't very good at handling IA's new challenges. That probably scares off a lot of prospective players, and nobody benefits from that. I would recommend that the difficulty of early encounters be tweaked a little, such that the challenges would keep pace with the player's levels and access to weapons. Perhaps the enemies could get stronger as the player collects Cromwell's new items.
After wading through many criticisms and complaints about IA, and hearing Sikret's defenses against most all of them, I think there might be some genuine problems that are driving prospective players away. Sikret has made it fairly clear that he wants people to play IA "the way it was meant to be played," to avoid the three C's (cheap, cheesy, and cheating), and learn new tactics for harder challenges. I think Sikret wants to learn to be more like a true "tactician," to get more skilled at playing the game, rather than relying on shortcuts as a means of overcoming the challenges that Sikret has spent so much time trying to make difficult. Sikret wants to make the player think if they are to succeed. His analysis of many criticisms of IA is that they are motivated by ignorance of the mod and frustration at their own lack of tactical skill rather than any real problem with the mod.
I do not think this is right, and I don't think it's helping the problem. I'll grant that a lot of people could stand to show IA more of the respect it deserves, but I don't think IA's critics reliance on "cheesy" tactics is the problem. It often isn't clear what tactics ARE considered acceptable in IA. Sikret has stressed that IA requires more tactical skill, but when stage manipulation, game engine tricks, hit-and-run strategies, attacking from beyond the enemy's field of vision, traps, and many standard D&D spells are removed from the game, it's not clear what Sikret expects beyond using IA kits and items and spells. I have often wondered how Sikret and other players manage typical encounters in IA. What spells do they use? What items? What did they sell? What did they hoard? How did they position their characters? Who do they target? How do they know which tactics work and which are doomed to fail? I think much of the frustration at IA is because the game is indeed fundamentally different from BG2, and they are left with almost no clues as to how to play this very new and very difficult game. I was bewildered by IA when I first started playing it. The enemy's new immunities and abilities appeared to come out of nowhere. Seeing a dragon in the Copper Coronet sewers, and getting killed by some mysterious spell just seconds after I entered the area, made IA appear nearly impossible. I have no doubt that surprises like that have done more to confuse and scare away players than to surprise and interest them. Merely repositioning the encounters so the easier fights are in familiar places would probably draw in many more people.
One might resist this idea on the grounds that it is counter to the spirit of IA, which is to provide a serious challenge for serious players. I like the idea of a strong challenge to the player's skills, but there is a great deal of marvelous content that many players appear to be losing out on. I stopped playing IA multiple times out of frustration with reloads or boredom from the somewhat standardized enemies of IA: the mages have SI and PFMW, plus immunity to area-effect spells by virtue of pre-cast spells that my Remove Magic spells could never dispel, and the fighters have damage resistances and regenerate. Invariably, the fights would become highly simplified. I would focus on surviving the enemy's spells and then gradually, slowly, eventually whittle down the monster after 5 minutes of fighting, not counting the initial period of tanking. I tried many methods and cheating and actually found them more satisfying than playing by the rules. Sikret has stressed the sense of satisfaction at overcoming a stronger enemy, and I've often felt that, but the seemingly endless stream of amazingly powerful minibosses--not just bosses, but even the Shadow Jailer and Vengeance Trolls shortly into a quest--made the fights feel more tiring and slow than challenging and epic. Having played the game for 10 years, I don't think it was lack of skill or experience that was holding me back, and having played many different parties and having tried many different mods and new tactics--I even did an Insane Solo Poverty with a Conjurer/Cleric and Kensai/Druid, instead of the usual sorceror--I don't think it was a reliance on cheesy methods that was holding me back, either. IA is hard. Hard enough to discourage people from taking the time to enjoy all the new content that IA introduces.
Making the early game easier would likely invite many more people to stick it through to the end, and I've gotten the impression that Sikret really does want people to play and enjoy this mod. One could argue I'm trying to make IA something it's not or defeat its purpose by weakening it, but there's much to enjoy in IA, and those Improved Crawlers and Noble Trolls and regenerating enemies of all stripes are probably the main reason why people don't finish IA, and walk away without appreciating what it has to offer.