Tanker solos Giant Monster - Film at 11!!

Cold/sonics and sonic/colds, mix and match. And yes, the hula hoop of doom is teh awesome.

Three colds FTW!

Assuming you can stand playin' one of dose tings....

I was wondering if I was being out of context but if you can reply to Slainsteel like this then I don't feel like I was.

I know that alot of people do enjoy team mixes that are randomly put together because there is more to a game than just speed. Speed teams are often the easiest teams to be in and taking out time. are practically challengeless.

From this thread I'm guessing that Ill/Rad and Ill/Colds are considered broken. Therefore Illusion is broken.

I used to have most the powersets all up on excel once but it was back in days before IOs. Nerdy days *shudders* . 'If' Slainsteel is completely right for the purpose of speed teams wherein survivability is not a worry I wonder what could be fixed with Tankers to close that small hole up for balance to allow Illusion to stay the way it is.

I will make an Ill/Cold just to understand them better.

Well, if they can't complete the TF, then they're not doing it fast or efficiently.

My point was that 'speed' and 'efficiency' are common goals in this (and most) game(s), not that everyone who attempts it reaches the same peaks as others.

Eh, I think your first reaction was more or less appropriate. Sleet can be double-stacked, but it takes awhile to get there -- and there is (or was, last I checked) a bug with the power that ends the duration of the debuff upon the rain's end if the target is still in the area.

Heat Loss can be used by Cold characters to stack another -30% RES, but that debuff is only available about 1/3rd of the time, best-case.

Then you've got Storm, which has a Sleet analogue and no other -RES. And Radiation, which can't double-stack Enervating Field. Then you've got Dark, which can double-stack Tar Patch pretty easily, but again it takes awhile to get there. Then you have Empathy, Force Field, and Kinetics, which have no -RES at all.

Pain Domination can (AFAIK) only stack about -20% RES. Thermal can only stack about 22%. So you're left with Traps, Trick Arrow, and Sonic Resonance.

It's certainly not a stretch to argue that the ability to stack 60+% in RES debuffs in a gameplay-relevant length of time is comparatively rare. Then again, if you're cherry picking every team, it's also not a stretch to argue that the -RES debuffers you recruit will have that ability.

Originally Posted by Iggy_Kamakaze Nice build

That's true, although its tricky to quantify how fast and effective that is since it relies on a team mate moving into an optimal position to affect all or most of the targets.

yeah, the best debuffer build I know of (I really can't stand playing the things, so there may be a better one out there) is a cold/sonic. I made a defender version (leveled it to 50 twice, STILL don't play it).

Here's the build I was gonna put on it if I ever decided to invest in the toon:

Hero Plan by Mids' Hero Designer 1.92
http://www.cohplanner.com/

Click this DataLink to open the build!



Yeah, sleet was hitting for 48-ish percent? And that was not even close to an optimum build.

Stacking debuffs are SO broken, but the Dev's seems commited, so whatever.

I'm just glad they gave Tanks one

Thanks, Devs!

That is correct, thanks, fixed

So when speeding TF's, we have two bottlenecks usually.
The first bottleneck is the AV, which is not really 'much' of a bottleneck since the AV goes down in 10-30s with these teams. But, since the argument was that in that particular case, a tank would help the team more because of the resistance debuff multiplying across the team.

The entire point of this exercise was the prove that the tank's lack of damage at it's damage cap, is not made up for by the -20% res that it provides; on a debuff heavy team, since as your debuff increases, that 20% is a smaller and smaller part of the final resistance debuff.




Actually this helps the point I was making more than detracts from it.
If you look at the further numbers, as debuff increases, the tank's inclusion starts to hurt more, quite a bit more.

Since not all debuffers will have their debuffs up, or some might miss, I often take 4 or 5 debuffers. That means that the tank is now competing even more so with the damage dealer.

If a tank comes, he's bringing 20% more debuff, and we're losing about half the damage of a scrapper. Now, half the damage of a scrapper is very relative term here - if the debuff is -100% resist, the difference is actual dps loss from the tank doubles. He still is providing 20% +damage (in effect) to the team, but the rest of the team is not very high damage, so his gains are limited.

Another reason for bringing 4-5 debuffers is also the fact that we are actually not fighting +3's on most the TF in quite a few TF's - the enemies range from level 50 to 54, with 54 only on the last mission.

On running my calculator again, the optimal number of debuffers changes to 4 and the tank loss increases to 7%, at 5 debuffers, it becomes 17%.

Secondly, we don't always get 2 kins, for the main reason that the other debuff toons aren't able to take advantage of fulcrum shift very well since they fire at range, so essentially with either 1 or 2 kins, the only toons that remain at their damage cap are the melee's.

That being the case, only the scrappers, brutes and tanks are at the damage cap, the debuffers are 'maybe' at 150-250% (my own debuffers often end up at around that much, I don't think I even hit my cap on my troller or fender except in rare cases, this is even with 2 kins). Running with 2 kins usually benefits us when we're running with multiple brutes, with scrappers, 2 kins don't get us much more speed than 1, sometimes even less.

So, in a more 'real world' scenario, the tank is losing half the dps of a scrapper, _and, is contributing 20% damage to a team where a majority of the damage is coming from just the Melee toons (considering debuffers already have lower base damage and aren't even at their damage cap).

Taking that into considering, let's just take approximate values here.
A kin does x base damage (considering the kin mostly just SB's, SP and FS's)
A debuffer does 2x base damage
A tank does 4x base damage
A scrapper does 8x base damage

With a team of 1 kin, 4 debuffers, 3 scrappers (the team that I've till now found to work the fastest. Of the records I've kept of our fastest runs, this has been close to the team composition on almost all those runs), we'd typically have,
debuffers at 100-150% damage buff, everyone else capped,
we're at,
x * 4 (kin) + 4 * 2x * 2.5 (debuffers) + 3 * 8x * 5 (scrappers)
= 144x damage

AV's realistically are really at about -150 to -200% resistance, even with 4 debuffers, we don't really get to 240%, like ever; by the time all the debuffs stack, the AV's are dead. Plus there are cases where Heat Loss hasn't recharged, or a debuff misses, etc.
So that being taken in, our total is,
144x * 2.5 = 360x

Now the team with a tank and 2 scrappers is at,
x * 4 (kin) + 4 * 2x * 2.5 (debuffers) + 2 * 8x * 5 (scrappers) + 4x * 4 (tank)
= 120x damage

Assuming bruising is on constantly, and shifts the AV's -resist from -150% to -170%,
our final damage is now at,
120x * 2.7 = 360x

I am giving the 'ideal' tank scenario here. Not all tanks are FA/(insert high dps tank secondary here). Further more, it's much harder to find a high dps tank than it is to find a high dps scrapper, and of course, bruising can also miss. In fact, unless the team's toHit is really high.

Keep in mind, I am giving ideal tank scenarios throughout my calculations. Finding high DPS tanks is hard - and when I do find one, I am usually a lot more sanguine towards taking the tank, simply because I know that the drop in speed won't be much. The rest comes down to the player playing the tank then; is he stuck in the tank mode of play or can he actually switch contexts and focus more on damage?

Given all these factors, again, why would I 'prefer' a tank over a scrapper or brute?

I should also mention, 'ideally', having more SoA's along helps a lot, except for the small problem is that most of the SoA's I see, aren't really built very well. A well built SoA with debuffs and high damage can replace the scrapper easily, maybe even a lower end debuffer (considering SoA debuffs last for less time and they have to shuffle between debuffs and damage, so one of the two does suffer).


I thought about for a while but it's more complicated than that :-\

I am not sure a 'correct' solution can really be reached without any algorithm that is not NP complete. The number of possibilities to cover can't be handled by just generating equations. The faster approach would be a greedy algo but for completeness, the best one can do is prune out descent trees that make no sense (like going over 5 kins say) to optimize the total number of equations to try.

Even then, realistically the algo would have to compute the team damage of the teams approximately 1.6mil times with different team compositions even if we were restricting AT's only to tank, debuff, kin, soa, scrapper and brute. To be more accurate, we really need to consider scrappers, fenders and controllers separately, which would put us at about 43mil equations - which would be fine for a modern computers considering the speeds we run at, but then comes the 'other problem'.

Considering real world factors, as you mentioned here, we need to run a partial simulation, rather than just use one equation.

My best guess would be to not use discreet ticks (since at a resolution of 100ms, the complexity of that code would be, err, horrendous), but rather run a thread for each AT, doing damage/modifying stats on the enemy, but there may be a better approach. Considering synchronization primitives and context switching time introduced because of that (1 context switch due to a sync primitive can take up to a 1000 cycles), this algorithm could take days to run even when written in C++. Assuming that we take only 0.01s to run a simulation, it would end up taking 4.98 days to run it just once.

So, basically I am not sure if we really can get a truly accurate answer for team composition unless we really simplify things, or, well, if we can find a really fast way to run the full simulation.

That last part, I am still thinking about.

Your current system uses only a single polynomial expansion for damage within a bounded space (eight maximum team members). By definition it will have a calculable global maxima and minima within that space, and so long as the relationships are all linear, the equations will either have a solution within the space or point to a solution along a boundary condition. For there to be no closed form calculation that solves for this, you would need to change your methodology entirely to something with either nonlinear or non-polynomially expandable expressions.

I'm not sure why you think it would take millions of equations though since all archetypes fundamentally follow the same game mechanical rules, just with different constants. Its worth noting that in spite of the fact that your code contains many different components representing the different archetypes, the entire program expresses just a single fundamental equation for calculating damage: essentially the equation I used above. In fact, when I last looked at damage mitigation discretely there were really only five fundamental equations in total expressed within the entire system (the tohit algorithm, the damage and resistance equation which includes heals, the +maxhealth mechanic, the mechanics of regeneration, and the mechanics of recharge).

Let's for a second forget about trying to find equivalent scrappers and tanks to compare; instead, my simple question is, instead of a Fire/SS tank, why wouldn't I rather pick a Fire/SS brute?

Slightly lower base damage, almost double the damage cap.
People's answers have basically been 'bruising!' which I believe does not help very much once you have enough -resistance stacked that the loss in damage exceeds the increase in damage by bruising for the entire team.


Um, I've already stated several times that on a majority of my runs, I am not anal about what everyone has to bring provided we have enough debuff/buff toons. The entire argument has been about are tanks going to actually make things better, or worse, even if by very slightly.

I've stated several times over, yes, bringing a tank along does not drop the speed by much, just that it does - everyone's point has been that it does not period. In fact, several people have been arguing that bringing a tank would 'increase' the overall speed of a speed TF versus an equivalent scrapper or brute.

If we were to just say that tanks usually will slow a speed TF down by slightly, not enough to really enforce a no-tank policy even - provided it's a tank built for damage versus optimal tanking? Then I'm with ya, completely.

If we were to agree that with damage heavy tanks, the tank WON'T slow down optimal speed TF, or will speed up the TF, then well, that doesn't make sense to me.

If we were to say that we cannot know either way, that also I don't agree with because I do believe that there is enough experimental and theoretical data that seems to clearly indicate that tanks do not increase team speed on such teams but often can slow it; if they're built for really high damage, then they would not slow it by much, or even really a noticeable difference.

...you really don't know many good high-damage tanks, do you.


Granted, that's not all that unusual. Damage tankers rarely bother with teams, so you'd not see many of them. They're usually busy farmin' up their billions.

On a team, a high-energy, high damage tanker is usually playing a 'spawn ahead', with the team mopping up behind. Soak alpha, fire AOE's, move on. The team can catch up if they're able.

I know when I go out and team with my fire/fire, I usually get a lot of startled commentary.

You're looking at this as a math problem, I am looking at it as a computer science problem.

Keeping everything constant, yeah, just math could determine the maxima; but if you want to start accounting for a non-static environment when buffs and attacks land at different points doing different amounts of damage to the AV based on what has and hasn't landed, how would one do that with an equation?

The reason I wrote my code with individual object instances for each toon was to be able to eventually add action objects of type attacks, debuff, buff, which then in turn can be used to run a simulation.

Maybe, we 'could' expand the simulation out into a linear polynomial equation, but I am not sure how that would work? Frankly, I don't know enough maths to be able to do that or to even know if it's possible, I'll defer to your superior knowledge of math to decide that.


If we do use the simulation as a series of steps applied by different objects from different threads, in that case, we would have to try every possible team combination (with optimizations as I mentioned to drop useless recursive descents), which brings us back to it being a NP complete problem.


P.S.: The advantage to doing it as a CS problem is that it can be used to calculate the math factors also along the way, without doing too much work

You're right, I don't know many high damage tanks; but I do know a few. When they run with us, trust me, they are still not staying a spawn ahead

In fact, the only time the entire team really gathers together is for AV's, otherwise most of the team is really split into solo, 2's or 3's at most.

But that conclusion is based on numbers that seem selected to derive that conclusion. For example, the value of a resistance debuff is actually based on the intrinsic damage output of the rest of the players on the team Swapping a tanker for a brute, for example, costs you the difference in damage between the brute and the tanker, but then adds about 20% of the total damage of the rest of the team.

Your calculations assume that "debuffers" have extremely low intrinsic damage. In fact, they assume the debuffers in question will have only 40% of the damage of scrappers (scrappers: 75, debuffers: 30). And that is base, not damage capped. Is that remotely likely if we pick, say, corruptors as our debuffers? And even if we pick defenders, that is a huge gap in damage when scrappers do not tend to have a lot of ranged AoE, even with epic powers. Defender damage mod is 0.65. Even assuming a 10% crit rate, if Scrappers are 75, defenders should be at least 39. Corruptors should be at least 45 even if we don't count Scourge, and we should. And what to make of your assumption that the base damage output of a Kinetics defender is only 27% that of a scrapper? (75 vs 20).

Those very low numbers reduce the benefit of swapping pure damage for lower damage and a debuff. And that's what makes your calculations not in fact giving the tanker the best possible case. In some of your assumptions you are counting the tanker contribution generously, but in others you're penalizing the tanker contribution severely.

I've counted lower numbers for debuffers and kins to account for the fact that for quite a bit of their time will be spent on casting debuffs/buffs. Playing a kin, between casting SP and FS over and over again, it leaves very little time for actually doing damage attacks. On my cold/sonic and ill/cold, my attacks are usually, sleet, benumb, heat-loss, 2-3 damage attacks, sleet, damage attacks. Most of the time, I don't even get to casting sleet the second time since the AV is dead already.

As I also mentioned, debuffers stay out of Melee and AoE range of the AV's, which often precludes them from FS/SP.

To further add to this, not every speed run has to be optimal unless you're trying to break records on *every* run, which really would ruin gameplay even for a speed junkie like me.

When we put together teams, we simply ask what people can bring and sometimes it's a tank; I've even run several teams with 2 tanks simply because people wanted to get their tanks for that run. Now if I get a team with 3 tanks, that's where I strongly suggest that 'someone' switch out to another AT :P

I don't think a tank is a bad toon to play, I rather enjoy the feeling of invulnerability I get on my stoney. I used to like my SD till I realized I was rarely logging him on, since most of my time IG is spent on speed TF's/trials.

FYI, on trials, I don't even bother with team composition past having 'some' debuff and preferably 1 kin per team.

Then you would go to a simulation, like I said you should if you reach that point. But your current system isn't there yet.

Simulations are not all that easy to make. People have mocked up things that were claimed to be simulations but they got caught up in the calculations: what they made were basically difference engines, not simulations. And to get proper statistical information out of them, they have to run fast enough to be able to do thousands, really millions of runs.

I keep telling myself I'm going to rewrite mine in C, but to be honest that's unlikely given the fact that I put so much algorithmic efficiency into the python one, it already runs billions of simulated seconds per hour. I should shoot it through 2to3 though, since I don't have a python2 on my system anymore.