Arcanaville

My memory says it did but my memory could be wrong. Let me check:

Looking back, in I13 when Shields went live Shield Charge was doing 1.7 Scale damage at its epicenter, and 0.7 damage across a 3 foot radius. Meaning, 2.4 Scale total at the center and 0.7 within three feet.


In I14 this was changed so that the power began doing 2.4 scale damage across its entire radius. This is actually when I believe the first "mistake" was introduced into the power. The power has a radius of 20 feet, but the two damage effects were constrained in the original power. The epicenter blast was obviously limited to just the center of the effect, but the 0.7 damage was *also* limited - to a 3 foot radius. It was doing 2.4 to the target, 0.7 splash damage to everything 3 feet away, and it was only doing *knockback* to everything else within its larger 20 foot radius. Now it was doing 2.4 damage across a huge radius.

But that could have been deliberate: I specifically recall players having issues hitting things with the power reliably because of the mechanics of the power. Someone may have deliberately removed the radius restriction to compensate. But if they did, in my opinion they went too high: they should have opened the radius up from 3 feet to something like 7 feet - because that's melee range, and the game seems able to allow melee to hit consistently from that range without problems.


In I16 SC was changed again, this time by giving each archetype its own pseudo-pet that dealt different damage (actually Scrappers always had their own version of the pet because their version didn't include taunt - but it did the same damage up to this point). The Brute version doing 2.4, the Tanker version doing 2.54, and the Scrapper version doing 3.6, which is I believe the version that exists today.


I could try to get the exact precise date of the changes, but I don't think that is necessary in this case. The history of Shield Charge, at least on live, seems to be:

December 2008 (I13): Shields Launches. SC does 2.4 at center, 0.7 within 3 feet, knockdown within 20 feet.

April 2009 (I14): SC changed to do 2.4 damage across entire 20 foot radius.

September 2009 (I16): SC changed to do variable AT damage: 2.4/2.54/3.6


To be honest, I had two completely separate independent chances to catch this issue myself, and now that I look carefully the reason I missed both times is because I was familiar with the power already and assumed the Brute version was representative of every version. In other words, had Brutes not come first in alphabetical order I would have caught this in October. Going to have to make a mental note never to make that mistake again.

That quote isn't intended to imply that art and craft are two separate things you should have some of, its intended to imply that art and craft are two ways of describing the same thing.

In other words, all of the art in this game is crafted, and it can be judged on the basis of its craftsmanship. All of the craft in this game is directed towards its artistic expression, and all craftsmanship in this game can be judged on its ability to deliver its intended artistic expression. There is no art in the game devoid of craftsmanship, and nothing is crafted that isn't a part of the art of the game.

You can't add more craftsmanship to this game, You can only *improve* the craftsmanship that already exists in it. And that is *always* a good thing. Similarly you can't take away the artistry of the game, you can only improve it or degrade it. But this MMO has exactly the same mix of artistry and craftsmanship as all other games: 100% of both.

Well, conversely I was like the seventh or so person to the level (40) cap when CoV came out. And that was on a MA/Nin stalker. If I want to step on the gas, my feet do reach the pedals. I just tend not to see the point.

The Standard Code Rant doesn't apply to Shield Charge mostly because it wasn't a programming error.

(The Standard Code Rant is basically an assertion that without knowing all of the relevant facts, attempting to extrapolate how easy or hard it is to add or modify a segment of code is doomed to failure. It specifically applies to programming moreso than in general because code interdependencies can make otherwise easy things difficult, and implementation details can make otherwise hard things easy. It doesn't usually apply to straight changes to the data which lack those entanglements. )


Actually, it appears to have gone live working as designed. The problem was that the design changes themselves went significantly beyond what they should have.

Is it possible for Castle to check every single change made to the powers system? Theoretically, yes. Practically, no. It takes longer to check a change completely than it does to actually make it. If Castle had to check all of the design and implementation details of every single change made to the powers system, it would be more efficient to fire the entire powers team and simply make all of the changes himself. You hope that the checking that is done is good enough that combined with Q&A it catches design errors like this, but it doesn't always.

An automated powers checking system would have a better chance of catching things like this, but it would require designing an expert system capable of understanding all of the various powers design rules and guidelines. There are a lot of them that factor into a design decision like this. I could probably write down over a hundred design rules that affect the design of individual powers, most of them "soft" rules (meaning: they aren't absolute, but require specific reasons for being violated).

In this case, though, it appears that whoever made the change did so deliberately, because they thought that was a reasonable change. And the number-space the devs operate in requires experience to guide intuition. I see an AoE with higher than Scale 2.0 damage, I already know something is wrong. I don't need to do calculations: for me that is an at-a-glance red flag that jumps off the page**. For whoever made the change, they may not have been dealing with the powers system for a long enough period of time to know that intuitively, and didn't see their calculations were heading to a bad place. It happens all the time, and I'm not immune to those kinds of decimal place errors myself (although usually I err in the other direction: I sometimes don't catch errors where the Scale value is too small at a glance).



** So much so that "3.6" is a bigger red flag to me than "200" is. I actually did the mental manipulation to realize that 200 was about three and a half in pet scale, and the three and a half is what triggered alarm bells, not the 200.

This is probably less a problem with the actual game engine, and more a limitation of the mission design tools. A trapped civilian under rubble is essentially a destructible object with a passive power that targets an entity with a power that forces it to play a lie-down animation on every pulse. Destroying the object causes it to immediately despawn and release the civilian from the prone animation.

Those enemies playing dead until attacked are in essence the same game mechanical thing as a trapped civilian, with just different release details (I am oversimplifying a bit the geometry issues associated with placing rubble over victims in ways that are not nonsensical and won't be oddly affected by the physics engine).


It may not even be a limitation of technology at all, and more the mission designers haven't gotten to making missions that require that sort of objective yet.

Its not quite that bad, because you only need to consider the case of a reasonably straight-forward solution to the mission, and then search for all possible ways to complete it quicker. All ways slower don't count.

Still, it would not be easy.


There's an orthogonal problem to allowing higher than standard rewards in AE missions, and that is the simple fact that there is no proportionality between "difficulty" and reward rates. There is only an implicit threshold of difficulty for a given reward rate which the situations must meet. In other words, even if it was possible to compute "difficulty" as a number, the appropriate reward rate for that mission isn't necessarily a function of that number.

The actual boundaries on reward rates for individual situations have as much to do with conditions outside of but related to those situations as within them. For example, the value of the reward that Hamidon generates has as much to do with how repeatable the trial is and how stackable the reward could be as it does how difficult the Hamidon trial actually is. Those types of variables are much more difficult to control for in the AE because people can easily publish many copies of the same missions with minor non-material (to rewards) changes.

Almost regardless of the difficulty of the arc, if the arc is indefinitely repeatable that already places a limit on the rewards that arc should generate. And that makes it difficult to allow more than trivial increases in reward rate above the standard one.

I can't think of very many kinds of objectives that couldn't be added to this game in much less time than rewriting it from scratch would take. Mission objectives are a part of the mission scripting system, and the mission scripting system is probably extensible in ways that might be time consuming, but likely non-disruptive to the rest of the game.

That's not to say they would be easy to add. Just that it would be easier to add than rewriting the entire game engine would be.

They play a really big part: they set it back several years.

PS: the last major numerical change to FA was a buff: a significant increase to Healing Flames.

I disagree with that strongly. Take the melee secondaries, especially the scrapper ones (including both versions of Invuln). They fiddled with them in basically every issue from release to Issue 5 by trial and error just to slowly converge on their Issue 5 values, and *then* they continued to fiddle with defense until I7.

However, knowing *only* what the numbers and mechanics were, the circa I7 performance of those sets could have been targeted computationally and generated that way right at launch. And in fact so much of what was done by trial and error in the past is now at least guided by computation, if not outright set by computation.

I personally cannot think of very many things that, were they to have been numerically analyzed for balance at the beginning of time with no prior experience in gameplay, the correct quantitative analysis wouldn't have generated far better results than the trial and error process did, and would have done it faster.


There seems to be a persistent, impossible to quash rumor that Castle employed some new balancing formula on EM. In fact, while I'm sure Castle had to consent to the changes, most of the EM changes were at least partially driven by BaB, because they were animation changes. In fact, the change to ET caused everyone to forget that literally days earlier we were discussing the change to Energy Punch that came about as a side effect of changing the punch animation itself, which many powers share (including the original target of the change: Dark Melee).

One of the things they were trying to do was eliminate the 0.67 cast time powers, because it was difficult to fit a variety of nice looking animations in such a short cast time. And energy punch's cast time increase had a similar impact on EM as the ET change did across a large range of recharge.

There is no formula that factors in cast time that affects EM. Only Claws and Widows have a formula that incorporates cast time, and its a very rough one that I wouldn't personally say incorporates the perfect set of balance requirements for such a formula.


In any case, to say that EM was "asymmetrically balanced" just because it did more single target damage and less AoE is missing the "balance" part of "asymmetrically balanced." That's just asymmetrical. The balance part would have ensured that its strength sufficiently counterbalanced its weakness relative to peer sets for the net overall value of the set to be the same to within the margin of error of the balancing system. There was (and is) no system that currently can make that statement about either version of EM.

Which is the problem in a nutshell.

My main took 908 hours to get to 50 (as reported by a civilian).


As a rule, I don't PL alts for the sake of PLing alts. I have been in PL situations before, always to observe or learn the technique, but since the alts I've done that with are also alts that I test other things with, often which debt-cap them continuously, it all evens out. Alts I roll specifically for the sake of playing the powerset combination or character I basically never PL. In fact, I still often go through the tutorial.


So on the PL front, I've probably *tested* every PL trick that exists, but I don't *use* PL to actually level alts. I'm pretty sure I'm not that rare of an anomaly in that regard.

I was responding to this:

The conversion is a fiction of reporting, so the exchange rate has no effect on the profitability of this game. All of the expenses are in US dollars, all of the revenue is in US dollars, and both are blind to the exchange rate. The exchange rate itself is just a static multiplier in the consolidated report.

It *is* important to factor in the exchange rate when attempting to extrapolate reported revenue (in Won) to likely subscriber rates, because changes in the exchange rate will change the Won-denominated revenue of those subscribers. But because we don't have subsidiary expense reports, we can't really compare expenses to revenue in any denomination. And if we did have those numbers, we wouldn't really need to know the conversion rate, because expenses would be in Won and revenue would be in Won, and we could still subtract and get a clean profit or loss number (of course, it would be in Won). But without those numbers, there is no "good news" aspect to the exchange rate, because there's no expenses number to apply it to at the moment.

In other words, what I'm saying is the exchange rate is neither good news nor bad news to profitability. It only obscures the true revenue numbers.

Anything that can import our data and have it work in a similar way isn't CoH2: its CoH 1.5. And there's no reason whatsoever to expend all the resources necessary to rewrite the game from the ground up just to get a CoH 1.5. I find it difficult to believe this would even be a viable option in jest.

If you want basically the same game as CoH 1.0 but with improvements, you just add them to the current game; cf: Ultra Mode. If you're going to start making an entirely new game, it will be an entirely new game.

Why you make a new game is to (hopefully) fix the mistakes of the past, and try new things that are incompatible with the current game. The old game keeps running for people who like those mistakes of the past and aren't interested in new things that are incompatible with the current game.


There's so many things that would basically wreck any attempt to seriously import us into a new game without essentially making up some arbitrary conversion (no different than attempting to recreate a CoH character in CO). I think its extremely unlikely the enhancement structure of CoH1 would survive into a sequel. It would probably be an invention system, but a totally different one. So slotting is gone. Without the exact same mechanics for endurance and recharge, many powers in many powersets would require major overhaul. I would expect any CoH sequel to have either linear or diminshing returns on things like damage mitigation and offense, analogous to (but algorithmicly different than) CO, which also radically changes many things. I suspect the entire Defense/Accuracy situation would be completely rethought, which would make sets like SR, EA, and FF require total overhaul.

And these are changes that are both highly likely *and* automatically torpedo character porting even if the devs thought character porting would even be a good idea. If they decide to start monkeying with powerset progression and power tiering, its all over.


You should probably assume that any game that looks like this one, only better, will actually *be* this one, only better, with game improvements over time. Any sequel they decide to make will be totally different, and mostly incompatible, because its those very incompatibilities that are the driving force for wanting to make a sequel in the first place. Heck, for all we know the devs are looking at those early Alphas and thinking they want a second chance to make that, which would be totally incompatible with this game just on its face.

Such as?

Well, snapping my fingers takes only about one second, so it wouldn't set the devs back very far (I said "if I could snap my fingers...").

On a more serious level, replacing the current implementation tools cold-turkey would be disruptive. But in situations like this, the best way to phase them in is to design the new tools to interoperate with the old ones, and supplement them. That's possible, and in fact in a certain sense that exists now, just in a very unofficial and unconventional way.


I disagree in two senses: I disagree they would constantly fiddle without some really strong cause, and I disagree that implementing a reasonably balanced game is a hopeless cause. It just requires the right numerical approach. Unfortunately, the only way to demonstrate that thesis would be to actually implement an entire game system with that philosophy and demonstrate that there is such a thing as good enough in balance terms, and that it has a reasonable chance of being reachable in a real live implementation. At the moment, there isn't one I can point to, unfortunately.


This presumes balance is synonymous with equality. Nothing could be further from the truth. Real game balance supports a greater amount of diversity than imbalance does, because real game balance allows you to hit what you aim for. The reason this is not obvious is that the real cost to being unable to balance effectively is that an entire range of possibilities is simply never attempted, because they cannot be controlled. There is an entire shadow of our game that would theoretically work just fine, but you and I will never see. We can only guess at what that loss actually is, but consider this: the cost in time and resources to bring just the defense sets from where they were at launch to where they were in I5 was quite possibly comparable to the effort required to add all the defense sets added to the game since that time. We could have had those *then* and an entirely new set of things now.

Now, as to the issue of whether good game balance requires exact equality, I bring you Exhibit One:



This game should be required study for every games design program on Earth. It should also be required study for every person that wants to claim that reasonable game balance is impossible without making everything identical.

According to a lot of MMO players out there, this game is an impossibility: an asymmetric, balanced game (in fact, an asymmetric balanced *PvP* game).


When I was in Seattle a few years ago I went to the museum of glass. On the wall was a quote that summarizes my thoughts on this perfectly. I'm going from memory, but I believe this is an exact quote:

There is no art without craftsmanship, and no craft without artistry.


In my mind, talking about the numbers in an MMO as if they were something separate from and less important than the "art" of the MMO is like saying there's no need to practice the piano when throwing bowling balls at the keyboard is just as artistic and equally acceptable.

Once or twice a year, I re-read *all* of the dev diaries from those days (I have them saved on my hard drive). I keep trying to put those diaries into the context of what I know of the evolution of the game, and every year I learn a little bit more about how the game works, and how it probably got to this point, and I get a little more out of those diaries than I did the last time I read them. As a student of game design, I find the evolution of this game, because I have more access to its design than any other, fascinating even if also sometimes disturbing.


Anyway, today's omfg quote of the day comes from Rick Daken in part seven of the original Cryptic Design Journal:


Ok, enough fun with that. To GoldenGirl: notice the word "skill" in the passage above. I believe that at least part of the original Alpha design of the game had skills in the sense of percent chance to accomplish something. In particular, weapon skills or attack skills would be the base chance to execute an attack - what we would today call base chance to hit.

In the hypothetical non-combat portions of City of Heroes which to the best of my knowledge never really saw the light of day (but which kept getting indirectly referenced as the SSOCS after launch), other skills would have involved a base chance to do something else: base chance to hack a computer, base chance to find a clue, base chance to pick a lock, etc. That sort of thing.

That's my best guess based on all available evidence, anyway. It seems logical given Rick Daken's comments about the inspiration for the original CoH rules system and game design.

Well, let me put it this way. I'm not a producer or a scheduler, so I don't know for sure if its possible. What I do know is that if I was in a position to do so, I would do everything in my power to make it possible.

The difference is, of course, I might fail. But it wouldn't be for lack of trying.

I was one of those people who strongly advocated that the AE launch without rewards myself. Having said that, I would rather try to fix the existing system than tossing it aside. While I'm not a fan of the reactive patching, I believe its possible to make the AE, if not completely unexploitable, then no more exploitable than the standard PvE content. It wouldn't be easy, but I've put enough thought into it to not be speaking purely hypothetically.

I can't think of any other way to judge the design tools as being "better" other than to be able to generate the same desired results faster and more error-free. Every balance problem Castle works on is a critter we don't get: every error correction translates to powersets we won't ever get to play. Think about where we would be today if just *half* of the balance problems and typographical errors from release to now were prevented by better design and analysis tools.

The more I think about it, the more I think I should just write some.

Alrighty then:

Two things:

1. It isn't only public companies that have that mentality. In the IT industry of which I have been a player for more than twenty years, *most* of the service companies I've seen have imploded after large bursts in growth, driven completely internally by the notion that expanding the business for the best possible work environment doesn't have the fatal flaw that all other moments in time won't be that optimal.

2. Conversely, public companies are accountable to shareholders, but usually in one of two ways: they are presumed to either be able to make money of which the shareholders get a cut (through shareholder dividends) or they are presumed to take that money and reinvest it in the company which should make the company more valuable (through shareholder capital gains). Both expectations are reasonable for investors. The problem isn't the push to make money, its the time horizons involved. And actually, it is rarely the rank and file shareholders that are pushing these get rich quick metrics, it is the management and the board which are, for their own personal purposes, usually incentivised purposes (hit this mark, get a bonus). And those people would be just as driven in privately held companies.


In fact, of the three major problems with the financial industry, this is one of them. There are many jobs in the financial industry in which you're allowed to take this gamble:

1. 90% chance everything fine, you win 5 million dollars.

2. 10% chance everything goes to hell, you lose your job.

This is an easy choice for a lot of people. The problem is two fold. First, the downside is far too low compared to the upside. In purely quantitative terms this is a good bet to make.

Second: most intelligent people know that's a lie, and the bet is really this:

1. 90% chance everything fine, you win 5 million dollars.

2. 9.999999% chance everything goes to hell, you lose your job, someone else loses a lot more.

3. 0.000001% chance you destroy a larger section of the planet than you have the authority to.

Supposedly intelligent people** round the last one off to zero and tell themselves that #2 has no innocent victims so they can sleep at night. The round-off error in particular is also one of the two root causes of the last financial system collapse (the other one is one statisticians might call the inverse gambler's fallacy).



** People keep calling the people who caused this problem "highly intelligent people" as if this is a given. This is at least somewhat of a lie: there are only two possibilities: these people were smart and took unacceptable risks deliberately, in which case they were evil, or these people did not understand what they were doing, in which case they were idiots. There's no in-between. It took me only a few minutes of research after the meltdown happened to be able to state mathematically what happened to the CDS market. It shouldn't take intelligent financial wizards very much more time than that.

Paragon Studios almost certainly operates in US dollars, and doesn't convert their cash into Won every night and ship it back to Korea. I doubt any "exchange" actually takes place for the most part: the money starts in US dollars and stays in US dollars within the Paragon Studios entity.

The NCSoft reports are consolidated financials which almost certainly follow the generally accepted practices for such things using the international rules for consolidated financial statements. These basically would cause NCSoft to report Paragon Studios (aka NCI) numbers by simply translating them, at the time of the reporting, into Won by either a cost-averaging method (unlikely) or as of report date (most likely).

You could take all of the numbers, look up the exchange rate as of the report date, and probably reverse engineer the actual numbers in US dollars to a very high degree of accuracy. There's almost certainly no other weird strangeness going on in the financials than that.


Frankly, the numbers are what they are. Even with the various sources for margin of error, the three quarter trend eliminates most of them as being anything but secondary modifiers to the general trend. I don't think they are cause for panic as such, but I also don't think the numerical trend is anything but real either.

Going Rogue may reverse that trend, but the important thing will be *how* it reverses that trend. I'm expecting a revenue spike regardless. But I will be looking carefully for substantive changes that persist over the following two quarters.


As to what this says about NCSoft's possible long-term commitment to City of Heroes, I think the cost estimate of about $94,000 (64,000 GBP) per employee is probably a bit high. And its probably the worst kept secret that PS is shuffling people around to work on a parallel development project for NCSoft so its difficult to say how NCSoft cost accounts for Paragon Studios. Having said that I suspect that short of a literal catastrophe, NCSoft would not pull the plug on City of Heroes while they were working on a possible successor MMO. I think there is a real danger of CoX losing critical mass for an MMO (there are those that would probably argue it already has) but I don't think we're there yet.

In my opinion, the biggest problem is that trends are powerful things and difficult to overcome with singular events. No matter how good Going Rogue is, to overcome a subscriber downward trend would, in my opinion, require a sustained counter-trend in content release. We have to be able to get to that target of one significant content release every quarter or so, in my opinion, and if the infrastructure doesn't allow for that, we have to change the infrastructure so that we can.

To be honest, if I was Brian Clayton, that would be my number one skunkworks priority: figure out a way to make tools that free the content developers to release content at least twice as fast as they do now. And I'm fully aware of what I'm saying when I say "twice." I think its entirely within the realm of possibility, but would probably require a special parallel effort to generate.

I'm not an expert on DAOC or WoW, but the problem in CoH is a combination of all of the above and then some.

Without getting into too much detail, AoEs in CoH are balanced around hitting very small numbers of targets - usually between two and three, depending on an AoE formula which to this day I'm not sure what the person who made it was thinking. In particular, for spherical AoEs the formula presumes that a spherical AoE will hit (or rather is balanced around hitting) 1 target, plus 1 target for every six and a half feet of radius. In other words, a spherical AoE 25 feet in diameter is balanced around hitting three things.

On top of that, CoH is not designed around fixed cooldowns which would limit the utility of such powers. CoH is designed around recharge, which is basically cooldowns you can pay to reduce. This means the benefit of AoEs can be increased to very high degrees by simply slotting or otherwise acquiring the right amount of recharge. Do other games allow you to *trivially* take the best damage output powers and increase their availability by 60% or more?

On top of that, CoH provides lots of ways for players to manipulate spawn density (which improves the efficiency of AoEs) and includes AoE control on top of AoE damage. So players can arbitrarily increase AoEs to their efficiency limits, and they can take away one of the few compensating downsides to attacking a huge number of things simultaneously (drawing unlimited aggro).

On top of *that* CoH provides ways to buy yourself out of the only other balance limiter the game attempts to enforce by default: endurance consumption. You can buy enough endurance to power AoEs to a far higher degree than they were designed for, and in fact many players assume that the ability to do so is a birthright.

So: the four things that might temper AoEs - suboptimal target placement, increased counter-attack, higher recurring costs, and cooldown limits - are all either under the control of the players or can be manipulated to a very high extent. Conversely, players can easily exceed the metrics that AoEs were designed to be balanced under, and they can do so in a way that multiplies their net value.


And there's the last problem, sort of a synergy problem, the one that no game designer should ever deliberately put in their games. You have to balance your game around your average players, but you also have to be cogniscent of what your very best players can do, and ensure that while they may exceed average performance, they do not exceed the limits of your game or vastly exceed the balance limits of your reward systems. So its generally a good idea to make sure that the things you place into your game are things that average people can get value X out of, and with increasing skill you can get up to Y value out of it, and Y isn't several orders of magnitude higher than X. You want average performance and best performance to fall within a manageable range. Fireball can hit one target, or 16. It can have 17s cycle time, or 12s cycle time or even 4.2s cycle time in theory. Its range of damage output value (before counting damage buffs and slotting ranges) is about a factor of 65.

Balancing for a theoretical range of 65 is just setting yourself up for failure.

Or AoE becomes something that requires a higher level of skill to leverage than now.

AoE is just plain broken in this game, and I'm presuming game designer observers have taken note to never, ever, ever let it happen again, either in potential sequels to this game, or any other game (but of course, that's a bad presumption).

The fact that there are such reference terms as "AoE sets" and "single target sets" is a symptom of the problem. All attack sets are single target sets with AoE attacks. However, the advantages of AoE are so high in many cases that a set with one more or less AoE than another can sometimes be classified as a totally different kind of set entirely. That's a design error. Just an extremely difficult to compensate for error, especially in a game where the players are addicted to AoE damage.

But that's why I'm on the fence on things like intelligent scatter. I believe the game would be better for it, but a lot of players are used to the way the game works now. Without a better reason than personal preference, I would be hesitant to make such a change to a long-established game.

I would have *zero* inhibition to make it mandatory in any new game.


(It wouldn't take much: the simplest solution that doesn't even require sophisticated AI would be to have every spawn have a designated radius, and have a random percentage of the critters in the spawn simply walk back and forth across the radius in random directions. This would prevent them from being clumped up all the time, but it would still allow players who wanted to leverage AoEs the opportunity to either wait for the best moment to use them, or attempt to force the critters into a smaller radius with things like knockback powers or other effects. That *alone* would do a lot to even out the single-target/AoE disparity that exists now.)

Very difficult in an MMO. Those AI scripts will be running one per critter, for every critter spawned in every zone and in very instanced mission. Very expensive in terms of CPU cycles.

There are clever tricks you can play. You can make a "critter AI" that is actually a form of hive mind, that controls an entire spawn in a way that simulates individual minds to reduce computational overhead. You can make triggered or reactionary AI that doesn't run a complete script, but assembles a set of triggers that are externally fired (as opposed to a giant if-then tree). But its still potentially very expensive to do on the server side, in terms of the CPU cycles the AI ends up burning.

Ironically, there is an "AI valley" similar to the "uncanny valley" of animation perception. Really simple "stupid" AI that is very straight-forward tends to be very difficult to game, because it simply doesn't react predictably or at all to player actions. Really smart AI can also be difficult to game. But in-between AI tends to be smart enough to be manipulated and dumb enough to be deceived. That "deception valley" is something very important to avoid in designing AI in the general case.

That would be cool if we could. Perhaps one day. The critical issue is making "smarter" and "dumber" AI that were computationally inexpensive: AI burns limited server cycles. The trick is to make the AI seem smart without having to actually be smart, and I think that is theoretically possible. But probably a low priority for the devs, since "smarter critters" isn't something that improves the game in a visible way to most existing and prospective players.

There are catches, though. Whenever I've brought up the subject of making smarter critters, one immediate thing that comes to mind is making them scatter more so that they don't all bunch up in stupidly obvious AoE bullseyes. But there are always players that respond that they don't *want* smarter critters: they want bowling pins they can just mindlessly vaporize with AoEs, and lots of them. Making smarter AI optional within the difficulty settings allows players to opt-out of that behavior, but it creates a conundrum for the devs as to whether that actually means the improved AI creates a leveling penalty on players that opt-in.

If we were talking about a hypothetical sequel to this game, where there was no need to honor pre-established expectations, I know where I stand on this subject. Its a little more grey when we're talking about making changes to the game that will monkey with people's difficulty expectations without a balance-significant reason for doing so.

Castle confirmed my description of the powers design tools in the last closed beta, as well as my description of its limitations: in particular, nothing happens in an automated fashion, and nothing ever has happened in an automated fashion to the actual powers definitions. Its all done literally by hand.