Revise heroism infusion
Reworked heroism from 1.6.0 hasn't been a particularly popular inscription and I'm not a fan of disincentivizing opening fights with it considering one shot and spike damage protection is meant to be a big component of the infusion. This version is more similar to how it was in 1.5, though slightly less powerful.
Made sure to be compatible with current saves, but some stuff is a bit oddly structured to do so.
I'm against this. Adding a source of power scaling to the inscription represents a nontrivial increase in damage to a large portion of the game, including NPCs, and we've had a lot of concerns about NPC damage in general. It also impacts the value of saves for a large portion of the game with ??? impacts. Using the inscription system for damage prebuffing is pretty boring, too.
Heroism needs some sort of revision but if were doing something like this it should at least be banned from NPCs. I'd also note NPC powers tend to be really crappy even on Insane/Madness so buffing them is likely to have a decent impact.
I don't see your opposition as founded in gameplay. You're speaking entirely based off of theory without checking numbers or testing the impacts of those numbers in-game. We discussed this some in PMs and as I mentioned, 1.5 heroism had in practice given effective powers with its stat boosts due to how weapon damage calculations worked and this was not an issue for NPC damage. These would be in the average case near triple the impact of the powers my version of heroism would give, sometimes even more if using a weapon with extra scaling off of stats.
If you don't believe me then let's run through some numbers for how much this would actually increase an enemy's damage. Consider the worst case scenario for endgame heroism buffing an enemy in High Peak on Insane: the average High Peak rare+ has around 100 in its highest stat, 150 if it gets in-class talents or gear boosting stats. We'll take 150 as the number going with the worst case scenario, and will also assume the heroism they get fulfills the random chance to align with their high stat. This would be a raw power boost of 150*0.25+10 = 48. If the enemy has the average power level of high peak enemies, around 65, this will raise powers from 65 to 65 + 48/4 = 77, or an 18% increase, which given the combatDamage formula would result in an increase in weapon damage of (77 + rescaled statmod)/(65 + rescaled statmod). Being generous and assuming the stat contributing to weapon damage is only 100 and not the 150 for heroism (which is unlikely to occur due to how randbosses prioritize stat increases but, again, worst case scenario), this would for a 1h result in (77 + 86.25)/(65 + 86.25) = 8% increase in damage. Call me brazen, but I'm not particularly concerned over a rare infusion giving a 3 to 5 turn temporary buff of 8% damage (usually closer to 5%) to enemies.
As for making saves unpredictible, the impact actually can be calculated for an estimate. The actual effective power increase tends to be between around 10-15% regardless of where in the game it shows up due to scaling based off of stats, which is also the primary way rares have their powers scale. This will represent early game an increase of about 4 to 5 effective power, midgame an increase of about 6 to 7 effective power, and lategame an increase of about 8 to 9 effective power. For actual numbers, this will be a 3 to 5 turn buff making enemies' debuffs about 10% more likely to land earlygame, 15% more likely midgame, and 20% more likely endgame, though I'm unconcerned with endgame and to an extent midgame given how prevalent immunities and cleanses are. I would also once again reiterate that heroism is a rather rare infusion and won't be showing up frequently, it will be closer to 1 in 5 at a high estimate.
As for your claim prebuffing is boring (which would not only be for damage and would in fact primarily be for one shot and burst protection, as I mentioned in the initial MR message) I disagree. This isn't a case where I can really give solid numbers to back myself up, but I will provide the anecdotal evidence that 1.5 heroism following the design my version does was much more popular of an inscription than 1.6 heroism, which discourages prebuffing. I would also make the claim that it is bad design to discourage prebuffing on an inscription intended to provide burst protection because without prebuffing it cannot help against one shots, but I recognize that this is arguable and based partially in personal preference.
I'm not sure what being "founded in gameplay" means here. Doing test runs over and over wouldn't be a good way to assess the impact of a behavior change to one inscription, and at a guess, you didn't mass games doing that either.
I based my comment off a basic principle of not wanting to inflate NPC offensive stats on a broad level in the current state of the game. At the time I wrote that you hadn't done any math either, which is fine, there doesn't need to be thorough math for every suggestion. Your suggested change is using the exact same "this seems good/bad" level analysis that my response does, and what most suggestions/criticisms use. Adding a bunch of math later and then attacking me for not having done such a thorough analysis doesn't seem fair. For the record I formerly had the word "strongly" in my post which came off as more aggressive and certain.
I haven't said anything about 1.5 Heroism, but I don't see anything you said about that I disagree with.
I'm not going to go through all of the numbers there, but that looks about right for a basic analysis. If the claim amounts to 4-10 effective power isn't enough to be meaningful on NPCs then I disagree. There might be ways of relevance that it gets stacked higher too, though not commonly. We can speculate (and yes, this is speculation, the same thing I did) about what the full impact would be and I suspect your analysis is about right and it wouldn't cause many problems.
That doesn't mean I have to think its a good idea to inflate NPC damage via inscriptions, nor that I have to think its an interesting or desirable change. I'd strongly prefer to change Heroism in a way that doesn't link it to damage, though I'm not quite sure how. Dropping the HP scaling part seems good, its mathy and unpopular. Refreshing on kill seems like a neat thematic addition. Ideally there would be at least some decision tension on when its activated (the HP scaling only incentivized waiting, didn't prevent using it at high life) but thats easier said than done so whatever.