Redlynne wrote:

Chicken ... or Egg?

It is a bit of a chicken/egg question. However, I disagree that questions like the ones I proposed are tangential to the question of enhancement mechanics, as the mechanics of the underlying systems influence how you would want enhancements to function.

Returning to resistance powers as an example again, suppose it is determined that a "baseline" assumption for a powerset is that a single resistance power will provide 60% resistance with 4 enhancements. One of the things I agree is neat about the system you've proposed is that you can achieve this result using two different sliders: the base resistance of the power and the power's enhancement multiplier. However, at the same time this inherent variability gives me some pause, because you always need to keep an eye on what happens when someone "overslots" the power. The numbers might look like one of the following:

A) Base res: 30%. Power multiplier: 1. Res with 4 enh: 60%. Res with 10 enh: 77.4%. Res with 13 enh: 84.1%.
B) Base res: 40%. Power multiplier: 0.5. Res with 4 enh: 60%. Res with 10 enh: 71.6%. Res with 13 enh: 76.1%.
C) Base res: 45%. Power multiplier: 0.333. Res with 4 enh: 60%. Res with 10 enh: 68.7%. Res with 13 enh: 72%.

One thing to note here is that the diminishing returns on enhancements under A aren't really diminishing, from a certain point of view. The 6 enhancements going from 4 to 10 mitigate 43.5% of the remaining damage, while the next 3 enhancements mitigate 42.1% of the remaining damage. With this setup the enhancements actually become increasingly powerful from a very valid perspective.

You might answer that that's the power of the system; you simply use option B or C for the power. In a sense I agree, but it's the same thing that gives me pause. If another powerset has less mitigation in other ways and allows for 70% resistance as a baseline, then you have to use different numbers. The numbers change again if you spread the resistance across multiple powers as you will probably want to consider the substantially higher number of slots that would have to be invested for peak performance. The problem gets trickier again if you introduce a pool power that can be added alongside both sets, as assigning the correct numbers to keep the bounds of functional resistance where you want them becomes very complex.

[I]Maybe[/I] this isn't a problem. Maybe the system allows enough control that it can be handled with elegance. However, my concern is that the numbers would have to be [I]very[/I] carefully assigned individually to each power to avoid too great a discrepancy in performance, generating a lot of complexity in both development and character building. It's a result of the interaction between the artifice of the square root function/power multiplier and the way resistance scales, and I expect it would necessitate making power multipliers on such powers very small. Otherwise, you could quite quickly end up with the kind of situation you described with CoX's defense sets pre- and post-ED occurring simultaneously in the game.

Whatever else you want to say about ED, at least the sharp drop made it possible to predict the upper bound of mitigation provided by a defense or resistance power. To some extent I think that's necessary, and to that end my instinct is to favour a function on the enhancements that has a horizontal asymptote in order to have a guaranteed upper bound regardless of how many enhancements are used, and further to ensure that the "baseline" number of enhancements (presumably 4) gets pretty close to that upper bound. Lessening the disparity between baseline and overslotting would allow a much more consistent use of numbers for power multipliers, which I think is a good thing.

Redlynne wrote:

Thank you for noticing that under my system, "perfecting" the optimization of a build is a challenge for the Human to solve, rather than for the spreadsheet to analyze and provide the "correct" answer for. So as far as THAT issue is concerned, I've built my structure properly in order to toss the optimization problem in the lap of the Player. PEBKAC problems are not my responsibility to solve. ^_~
Pyromantic wrote:
The result would be higher (potentially much higher) damage in the hands of someone with the math skills or (more likely) an external tool to crunch the numbers. I think great care would need to be taken to avoid the accidental creation of the have/have-not divide you are working to avoid, even though that divide is gated by math instead of some other factor.
Hence why a Build Planner Utility would be a wise thing to do, and if released as an App could potentially drive interest in the game. The simple fact of the matter is that people who PLAN what they're doing will often be better off than people who do things haphazardly. That's just the way the world works. Even with the game of tic-tac-toe, people who know "more" about the game will be better armed and better prepared to play it than those who have done less (or zero) analysis of what works or what strategies to use in order to be (more) successful.
Besides, ANY game that involves numbers in some way is going to skew an advantage towards the people who learn the math involved. It was true for Defense Powersets in City of Heroes, and understanding what the Defensive Softcap was (and how to build in order to reach it), and it'll be fundamentally true for whatever happens in City of Titans. That's just how the world works.

Well, optimizing a given number of slots for, say, damage, under post-ED, pre-invention CoX, is about as simple as the question will ever get. It's pretty much just add damage to the power contributing the most under your current attack chain until you hit 3 SOs, then move to the next best power. Even limiting ourselves to this and ignoring things like working out ideal attack chains, the problem has gotten quite a bit more complex, but it ultimately is still just a math problem with a correct solution. The solution won't be as simple as just slotting your big hitters until you run out of slots, or keeping the number of slots balanced, and I think we'll find a planner absolutely vital. I don't know what percentage of CoX players routinely used a planner, but I certainly knew some players that had no interest in the level of tinkering with builds and were perfectly happy with the guideline of "slot to 3, then stop."

Of course, system mastery always will (and should be) rewarded. I'm not advocating the removal of that. In many ways I'd enjoy this level of complexity, as I trust my own math skills (and my satisfaction to invest the necessary time) to do a decent job of optimizing builds under this kind of system. My instinct is, however, that unless you keep the power multipliers very low, the disparity between a heavy optimizer and an "eyeball it" player will be dramatic, and we might be rewarding system mastery too much. It's just really hard to say without seeing what a powerset might look like, but this further pushes me towards an asymptotic approach to diminishing returns on enhancements.

Redlynne wrote:

A substantial reason for doing it this way is because ... it's what City of Heroes did, and is therefore "familiar" to anyone who used TO/DO/SO/HO Enhancements and knew how to combine them (very few people did after Inventions came along).

Fair enough. I just don't see that as a good enough reason. I don't think anyone got particularly excited about finding SOs as loot in CoX unless they were under level 25, which is no longer a consideration under your proposal. If you go far enough in the system as to make enhancements purely on/off (with no change in effectiveness from relative level) and make drops frequent enough that maintaining the build is fairly trivial, then it makes sense to me to take the extra step and remove basic enhancements as loot altogether. I don't see that anything fundamental to the experience has been lost.

I expect there's more I'd like to ponder and respond to, but that's it for me tonight. :)

Redlynne wrote:

Under my system, hyper-specialization is perfectly possible ... but it's up to the individual Players to decide for themselves if that is WISE. I'd argue that hyper-specializing in a way that makes your character ill-suited for much of the game would not be a wise decision ... but I'd expect the Players to be "smart enough" to figure that out for themselves, rather than relying on building a "Stupid Stopper" into the underlying, fundamental structure of the game itself that FORCES them to "make the right decision" because *I* as the Game Developer "know what's best for them" and have decided how they *ought* to play the game (ie. MY WAY, as opposed to their way).
There is no such thing as Fool Proof ... there is only Fool Resistant.

Thinking about this further, my issue is that hyper-specialization may not be unwise at all; it may be gamebreaking in a very undesirable fashion. There are two ways I could see this happening:

You might be able to specialize towards a very specific enemy group or other piece of content. I'm reminded of a tactic some people used in CoX with capped fire resistance (using fire armour) and high fire defense (using mostly sets) going into AE against solely fire-damaging enemies. Potentially, this could result in an extreme degree of efficiency with that content such that you are heavily encouraged to farm it over and over. I don't care if someone wants to farm, but we need to be careful that an extremely narrow focus in character build and/or content played isn't overly encouraged, or worse yet "necessary" to keep up.

Similarly, you might be able to, for example, massively specialize towards mitigation at the expense of damage. Without care such a character might mitigate damage down to a small fraction of what a more balanced tanker receives. This would make it practically impossible to design content suitable to both approaches.

Point being, I don't feel care here is necessary as a "stupid stopper" but to ensure that a variety of content is suitable to a variety of builds in both challenge and reward, something that I believe was fundamental to the CoX experience.

Pyromantic wrote:

Lin Chiao Feng wrote:
No, the ED curve is ham-handed and arbitrary. Exponential decay does a far better job if you need an asymptote, logarithmic growth if not, and inverse parabola if you need it front-loaded.

I feel the need to clarify here, in that you begun with the word "No" as if what you're saying disagrees with what I said, and I don't believe it does. I'm not saying the ED curve was ideal for CoX or would be ideal for CoT. I'm just saying that--while it had plenty of negatives--one benefit of it is that having a relatively predictable upper bound for the effectiveness of mitigation powers makes balancing an easier proposition. Easier, though I don't think it gets particularly close to easy. However, I am absolutely in agreement that another approach is better for implementing a predictable upper bound on enhanced values, should such a need be confirmed.

I said "No" because I think you're giving ED too much credit.It was for all practical purposes a brick-wall limit, a cliff instead of a rolloff, at an arbitrary point around 3 SOs. That's all.

Pyromantic wrote:

Using an exponential decay function is certainly an option. I think it's important that we draw a distinction between different mechanical quantities here. For example, flat increases in damage output inherently produce diminishing returns in a sense, while flat increases in resistance (under the CoX model) could much more accurately be described as producing growing returns. For this reason an unbounded function might be perfectly acceptable for calculating enhanced damage, while needing a bounded function to calculate enhanced resistance.

Definitely. However, my assumption in promoting the unbounded log curve was that there was another mechanism, outside the scope of determining raw enhancement bonuses, that would provide such limiting. A later step in the math, if you will.

For the record, I'm no fan of how to-hit, accuracy, and defense related to each other in CoH, especially dealing with order of operations and when caps were applied. I think it could have been done much better.

Which is pretty much where you wound up as well. As for your inverse-proportional curve, the only problem I have is that it might turn out to be too "sharp" of a curve if the coefficients are not well-chosen. An exponential decay would work well here.

That said, I think the equations you're working with need to be redefined in terms of:
[list=1]
[*]Unmodified effectiveness.
[*]Enhancement scaling factor: 100% for no enhancements, grows.
[*]Buffs applied
[*]Debuffs applied
[/list]

Things like to-hit will be more complex since there are more factors to include.

Power Play wrote:

Everything is SO complicated. :) What COULD be done is just allow all powers to come with six (or however many) slots already open on them, so the players can customize their characters' powers the way they want from the beginning. That way every new energy blaster won't fire just as fast and just as accurately as every other. Those who want faster, lower damage can slot more recharge and less damage. Those who want to be sharpshooters can slot more accuracy...and everyone can do that from the beginning. That would also keep characters from having to be given "free accuracy" from levels one through 10 as was done on CoH, where your character didn't have to start slotting accuracy until they hit level 11. That would make it easier on the programmers and simpler for the players...especially NEW players.

Why not split the difference? Assign three free slots per power when the power is selected. Other slots can be assigned as they are earned leveling up. This requires the players to make choices as to where the free slots go but they will be in no way gimped early on.

Redlynne wrote:

That's nothing more than straight up Min/Max Content Design by Players with intent to FARM. Such an occurrence is essentially ONLY possible in a Mission Architect environment, where the Players get to PICK what opposition they're going to be facing, and they Min/Max that Player Created Content to the maximum extent with every intention of Farming it until the cows run dry. Odds that the World Content Creation Devs for the game would do the same thing? Practically nil.
So when you say you don't care if someone wants to farm ... I'm going to have to take you at your (stated) word and say that this kind of worry is basically a non-problem for the game at large. What you describe is functionally an exploit because it's essentially a Min/Max For $DEITY Mode and is really only possible (let alone, *likely* to happen) when Players get to create their own Content to Farm.

There are several problems I see with this assessment. The first is dismissing player-generated content, which I strongly suspect will be in the game and will be intended to represent a significant portion of the available content by some point (if not at launch.) I base this on the fact that as a volunteer-created game it is more likely to rely upon the players for development, and that IIRC the devs have already suggested the tools they are using to generate content will be made available in the same vein as the AE designer. That kind of misses the point though; if players are in a position to hyper-specialize against certain types of enemies then they don't necessarily need to have design control over the enemies to achieve that end, despite your claim. [I]If[/I] players are given too much leeway to build their characters towards the content instead of building the content towards their characters the way AE allowed, then the potential is there for that kind of min/maxing regardless.

I'm not sure exactly what point you're making with your anecdote about slotting for defense first with a SR scrapper. Since you're talking about CoX it's much more accurate to describe that as an anecdote about being required to invest heavily before hitting the [I]expected[/I] level of performance, while I'm talking about the potential to push [I]beyond[/I] the expected level of performance.

You say I'm working under the assumption that nothing is given up to achieve a high level of performance but I can assure you that is not the case, as I think is clear from the fact I'm discussing hyper-specialization in the first place. However, I do think you are working under the erroneous assumption that because you have placed diminishing returns on enhancement value that pushing beyond a certain point (a point that lies within the structure you are proposing) is necessarily a bad idea.

Suppose for a moment that CoT has something roughly equivalent to the fire/fire or fire/mental blaster, an AoE monster. I could imagine someone getting the maximum number of slots in AoE powers and devoting practically every slot to damage, presumably after a sufficient amount of accuracy. Is this necessarily a bad idea just because you introduce diminishing returns on enhancement? Not at all. Even with "diminishing" returns the extra damage can make a substantial difference to the alpha strike, enough that such a character could have a significant edge in wiping groups before ever taking return fire. Is this potential gamebreaking?

I suspect the answer to that last question is "probably not," though it's not a slam dunk. The reason I lean that way is because of the nature of enhancing damage. What I'm really getting at here is that using a function like square root in the way you have ensures that the "distance" in the sense of absolute difference diminishes as the number of enhancements increases, but it doesn't ensure that the "distance" in terms of game benefit so diminishes. Your insistence that the diminishing returns you provide acts in a DISCOURAGING way to prevent overslotting past a certain point doesn't become true just because you write it in CAPITALS. We can actually see this in CoX; there was a modest diminishing return kicking in for the third SO in a power, but nobody suggested stopping at 2 slots for important aspects because the diminishment was relatively minor. The diminished return on the enhancement hadn't yet overtaken the benefit of continuing to slot. Of course, past the third SO the diminishment was so dramatic that the benefit dwindled down to almost nothing. However, you can't just assume that under any proposal you make this is going to occur somewhere within the spectrum you allow, or even that it will occur at all.

In the case of damage, those kind of diminishing returns are somewhat built in, because the "game benefit distance" between 10 damage and 20 damage is typically greater than the "game benefit distance" between 90 damage and 100 damage. That's why my concern at this stage isn't centred on damage, and you'll notice that I didn't talk about glass cannon characters much.

In the case of mitigation under a system resembling CoX's the opposite tends to occur; the "game benefit distance" between 10% resistance and 20% resistance is going to be dramatically smaller than the "game benefit distance" between 90% resistance and 100% resistance. As such, even if you apply a diminishing return function on enhancement that is no guarantee at all that the tangible benefit of enhancements diminishes. The result, I suspect, is that you would have to be very careful in balancing mitigation powers to prevent characters from being able to slot to the point that tank-like characters take a small fraction of the incoming damage compared to "expected" slotting levels. I alluded to a couple of the problems that I think would follow. One could be characters slotting mitigation powers that are extremely potent in a narrow range of content while maintaining offensive functionality, allowing a player to far exceed expected performance within that content. Another potential problem is that tanks can devote so much to mitigation that damage threatening to a "baseline" build becomes negligible, which would make it impossible to design content like the STF that is suitable to both.

In the case of mitigation powers the only way I see to avoid this under your proposal as presented is to [I]very carefully[/I] assign power multipliers individually, a process that would become more difficult as you spread mitigation across multiple powers, especially if those powers reside within pools available to all builds. As such I think it makes much more sense to use a structure on such powers that takes a top-down approach, ensuring that one cannot push beyond certain bounds no matter how much enhancement is applied. Working under the assumption that we want a model close to CoX, and since you are already using the enhancement system itself to apply diminishing returns, I see no reason that this limiting structure should exist anywhere other than in the enhancement system.

Pyromantic wrote:

The point is that in having that brick wall you can at least easily predict the approximate upper bound of build functionality.

Nothing approximate about it. It was a very well-defined, arbitrary limit. This isn't trade-off; it's game designer's fiat.

Pyromantic wrote:

It may not be particularly elegant but it does make the process of balancing power sets easier.

It does nothing of the sort. There was no per-power scaling of damage enhancements, for example. First at-level SO did the same thing no matter what power you slotted it in. Thus any adjustment does the same thing to any of those powers.

Pyromantic wrote:

Obviously I am not privy to the internal decision-making process that led to the introduction of ED, but I wouldn’t be at all surprised to learn that a smoother curve was considered and rejected because it was too much of an undertaking to refine it well enough to suit their purpose.

Well, we can all have our little fantasies, I guess. I think you give them far too much credit. ED reeks of "anything more detailed made the PHB's eyes glaze over, so this is what we got."

Pyromantic wrote:

In my previous post I mentioned that I think the enhancement system itself should take care of any decay that needs to be applied.

That's the fundamental assumption of the whole thread.

Pyromantic wrote:

I think any later step in the math of determining something like damage resistance should only be used in the case of (de)buffs or stacking powers, and that enhancement of such powers be self-limiting.

Yep. That's a later step than "how much boost do my enhancements do".

Pyromantic wrote:

I don’t see any need for another step that would only increase the complexity of the math and make the system less comprehensible to the average player. In so doing you could at least use the UI to tell the player “the next enhancement you slot will raise the power’s effect to this level” without worrying about any further functions acting on the effect.

The assumption is that the UI [i]will[/i] do those things. How cruel do you think I am? The game is on a bloody [i]computer[/i] for crying out loud, why would anyone force the player to crank up spreadsheets and stats tools and regression analysis and...

... Well, okay, Statesman actually [b]did[/b] believe in hiding all the math from players, which is why the stat readouts, including the "slotting this will do this to your power" popup, didn't happen until years after he was gone and Positron took over. And I'm supposed to assume great wisdom on Statesman's part in the design of the ED curve. Cognitive dissonance...

BTW, there's also an effort to get them to include a Mids-style build planner in this game at some point.

Pyromantic wrote:

As for dealing with (de)buffs I didn’t really think we’d hit that point in the conversation.

Nope, that was a ways off. Hence the roughness of the ideas. And I don't think you'd want anything at that late a point in the math to act on the enhancement instead of the total power effect, because that creates openings for all kinds of oddball bugs, and could even create a disincentive to put enhancements in a power. It's like not wanting to get stronger because you'd have to buy a bigger shirt.

Redlynne wrote:

Pyromantic wrote:
It strikes me that there is some fundamental gap in communication between us here.
That fundamental gap in communication basically boils down to something like this ...
Pyromaniac: "Cart First! Horse after."
Redlynne: "No no no, you've got it all wrong. Horse first, cart after."
Lin Chiao Feng: "Pay attention to how you hitch the horse and cart together and all will become clear."
{wash, rinse, repeat}
The fact that no matter what gets said or what clarifying statements are conveyed, you still insist on putting the Cart Before The Horse™ leads me to the inevitable conclusion that further discussion will not enlighten you (or Wyvern, for that matter) on these issues, and would therefore be a wasted effort.
/em walks away

The only conclusion I can draw from this is that further conversation [I]would[/I] be wasted effort. I find it very insulting that you work under the assumption you are here to enlighten anyone disagreeing with you, instead of considering the possibility that just maybe what you've proposed isn't perfect.

The irony of this is that a lot of the structure you used I agree with, but I had concerns that I think would need to be addressed to see it implemented.

The most fascinating thing about all this "Big Brain" arguing (Sorry, no insult intended!) is that the whole point of the game is to HAVE FUN!

Fun on day 1. Fun on day 100. Fun on day 1000!

Redlynne's entire proposition (from my point of view) is not about Min/Max building, but her perception about how players might have more fun at early-level play and more challenge at later-level play.

I was in CoX from pre-i2 (wasn't beta) to the bitter end.

My initial character was hard to play, but still fun for me because I was playing a Hero, who someday would be able to fly! And who could blast his enemies into rivers and over the edge of bridges!

When they started making early-play easier (by making Fitness inherent and giving travel powers early) at first I was "But what about all the effort I put in to get all this stuff!" -- but I quickly got over it. My alts started coming even faster then, trying more builds and crazier themes, since it was easier to get them moving around the city. In the long run, this stuff was great for me, so when I saw something similar happen in Lord of the Rings Online, I didn't feel the jealousy.

But before the "easier start", I was there for all the complains from Fire Tanks for their nerf. I understood why it was done, but I wasn't a strong Tank player anyway, so I somewhat ignored it. When nerfs hit my precious blaster, I wined for a moment, but it wasn't too bad, so I lived with the pain. (Later, when the class specialties came out and I could still fight when mez'd, you can bet I was crowing about that!)

As nerfs continued to be rolled out, chat started to focus on the PvP aspect of the game being developed, and everyone started to wonder what would happen when the data from Arena fights was collected and discussed. Would there be more nerfs to character? Would some get hurt worse than others? Would characters be playable in PvE if people built with PvP in mind?

Then the PvP nerf bomb (Lovingly called Enhancement Dysfunction by our thread host) hit, sending a shock wave through the game's community. Whatever else you've heard, whatever else you believe, whatever else the devs said at the time, I firmly believe that ED was a product of the data collected in the Arena and the fact that only a few character concepts were dominating play there.

Not that the WHY matters in the long run. The result was a Nerf to characters like I've never seen, before or since, in any other game I've ever participated in. It took a LONG time for the game to be fun for me again ... but eventually I got over even the ED debacle, although I know many people never did.

Why am I bringing all this up? Because every LONG argument I've seen in this thread, those attacking and some defending Redlynne's proposal, IMHO, misses the point of her proposal quite badly.

Her proposal, complete with its high-impact-start square root curve, is about a stable, FUN starting point for players, one that gets more challenging as the game moves on. One that can become MORE FLEXIBLE as the game develops, and one that is ENTIRELY based on the idea that this game is BARELY OUT OF THE OVARIES YET! (sorry if that's too crude).

Unlike Lin, I do not know Redlynne personally, so take whatever I'm saying with a grain of salt ... but if I were her, and the same people kept telling me over and over that my extremely detailed and clearly well-thought-out idea for a FOUNDATION for an enhancement system wasn't good enough because it didn't factor in quantum chaos logarithmic hyper-reality theory (and there were hints that my theory was a wave washing up against the breakwall of already-developed ideas), I'd want to walk away too.

Hang in there Redlynne.

Pyromantic wrote:

The only conclusion I can draw from this is that further conversation would be wasted effort. I find it very insulting that you work under the assumption you are here to enlighten anyone disagreeing with you, instead of considering the possibility that just maybe what you've proposed isn't perfect.

Of course it isn't perfect. It will never be perfect. And why shouldn't enlightenment be the goal? This is [b]Redlynne's[/b] proposal, and Redlynne will take the bouquets or brickbats for it.

If you feel you have a better proposal, [b]post a thread with that proposal.[/b]

[b]There is no reason to accuse Redlynne of bad faith for refusing to be at your beck and call.[/b]

Pyromantic wrote:

The irony of this is that a lot of the structure you used I agree with, but I had concerns that I think would need to be addressed to see it implemented.

Congratulations, you've just defined "concern trolling."

None of this was on any kind of implementation track. Ever. Redlynne has previously noted the complete and utter lack of dev input, even though the required IP releases have been turned in long ago. Whereas on several other threads, devs have mentioned that they're interested in what's in a thread or following it. That's a huge, huge sign that they have absolutely zero interest in this.

So try not to take this so personally, okay? I'm not getting these hours of my life back either.

This thing is dead.

*tries hard not to imagine what a nun of the "Scary Devil Monastery" does for prayer ...*

*again laments the lack of a "like" button*

I'm still giving this whole concept some thought, and while I'm not sure I agree with this sort of plan, I certainly can't dismiss it entirely. And the thought that the Global slots are overpowered, well... if this takes the place of a large degree of the set bonuses and incarnate abilities we had in CoH, then, as powerful as the global slots undoubtedly are, they aren't completely over-balanced.

My final WarShade build was running around with +24.5% Damage (all types), +109% Acc, and +141.3% Recharge, just from set bonuses. Add in a minimum of 30% recharge enhancement from the Tier 4 Alpha slot (to powers that were already at the ED cap for recharge, up to 45% for other powers) and Hasten plus a couple procs for more recharge. As a practical matter, call it @250% global recharge, PLUS whatever was slotted in the powers. Perma-hasten, Perma-eclipse, at least 3 puff-ball-of-doom out at a time, etc.

The Global slots (at level 50 at least) would have a hard time getting you to that level of recharge, absent some sort of set bonus scheme and/or incarnate-type powers. From what I understand Redlynne is expressly proposing this system with the understanding that there NOT be said bonuses, so the strength of the Global slots isn't necessarily overwhelming.

The concerns over resistance or other protection powers stacking, misses an interesting distinction I've not seen anyone bring up: the reason protection powers are a problem over attack powers is enhancements in them have a "global" effect by their very nature.

What do I mean?

Well, if you enhance an attack, you've enhanced...well, that attack, and only that attack.

When you enhanced, say temporary invulnerablity, you're enhancing the protection afforded your whole character. Perhaps a better way of thinking about it is that a protection power is in effect an enhancement added to an inherent "resistance" or "defense" power. Thus, an ehancement added to that is an enhancement of an enhancement, and this is why stacking suddenly becomes an important question.

But that specific question is part of a more general one, "how do we handle buffing a buff?"

At which point it's probably safe to say that we could apply a similar diminishing returns curve to buffs themselves (for reference, I prefer the yellow-ish curve from the graphs Lin posted. I like the idea of an upper bound, and that had the most natural feel to me of the exponentials).

So an enhancement in a buff power (including protection powers under that heading) enhances how much benefit that power gives you, but the total benefit you can get from all powers is also then affected by diminishing returns.

And naturally this curve would /also/ affect any external buffs too (so buffing the defense of a squishy would have a much greater effect than buffing a tank).

This gives the player the choice of max slotting a few powers, or using the free slot on several powers combined with one or two global slots, depending on where else they want to use their slots.

And it means no matter what options they later add, we know the effective max values anything can ever reach, no matter how people they fulcrum shift, or how rad/rad defenders you know, etc (yes, I know those won't exist in CoT. But there will be equivalents).

From what I've read, I think /that/ is an overall goal we can agree on.

I'm well aware that was your original intent. However, like it or not, those other elements were brought in too.

I was attempting to more clearly delineate the differences, in the perhaps vain hope of demonstrating that it is kinda both on-and-off topic in a weird way.

That is, enhancing an attack is self-contained and so can be considered in isolation. For these, yes, your original suggestion is complete and standalone.

Enhancing a defense is by it's very nature a globally interacting power, and so while perhaps outside the scope of your proposal, the effects of enhancing it almost require considering the knock-on, combinatorial effects to be able judge the scheme's reasonableness, even at this early stage.

Since this has come up again...

I want to be clear that my concern over resistance isn't based on the way resistance powers might stack; it's based on the underlying behaviour of a characteristic like resistance. However, power stacking would exacerbate the problem. That's why it isn't something that can be dealt with after the fact by looking at power stacking.

I took a quick-and-dirty approach to this earlier, but something more rigorous may shed more light. When you look at the value of an enhancement, it isn't all that telling to say "this enhancement adds 50%; that one adds 25%" by itself. The question depends on how you value the change in the effect that is being enhanced. How does that happen in the case of resistance? It's a bit of a simplification, but one way to measure a final resistance value is to think of it as a multiplier on your hit points. For example, if you have 75% resistance, you effectively have 4 times as many hit points. The function you would use for this is 1/(1-x). (So, in our example, the value of 75% resistance is 1/(1-0.75)=4.)

There are some things to note about this function. Firstly, it's well-defined on the interval [0,1). (Actually it could be used for negative resistance values as well, but I'll drop that as outside the scope of discussion for now). Secondly, it's increasing on that interval, which obviously it should be if it's measuring the value of resistance. Finally, it has a vertical asymptote at x=1 going to positive infinity on the left. This last point is particularly important.

Now, if we want to consider the actual value of enhancing a power (even in isolation), we can use a composition of functions. So if n enhancements are applied to a resistance power with base value R and power multiplier P using Redlynne's model then the value of the enhanced power is 1/(1-R(1+sqrt(0.25n)*P)). Assuming that R and P are positive values the unbounded nature of the square root function ensures that there will be a value of n at which you continue to have this vertical asymptote. In turn that ensures that the practical value of the resistance approaches infinity for a finite number of enhancements, and so the shape of the curve is concave up. Note that it doesn't matter whether or not it's possible to get that many enhancements under the system; the fact that the curve is concave up means that the practical value of enhancements is increasing as you add more, contradicting the notion that the returns are diminishing.

You might argue that it's only because of the particular way I'm valuing resistance, but I think any reasonable measure of resistance will have similar properties; i.e. it will be an increasing function with a vertical asymptote at 100% resistance, since at that point you would be impossible to kill with damage. The real issue is that the underlying mechanic places an infinite value on a finite value of the attribute. In such a case, using an unbounded curve to apply enhancements is problematic because the practical value of those enhancements won't be diminishing. The use of a power multiplier doesn't change that fact because the enhancement curve is still unbounded, having just undergone a vertical compression.

The use of a bounded curve corrects this issue. ED accomplished this by using an effective cap on the number of enhancements that could be used on a particular attribute, but I've suggested a smooth bounded curve such as arctan. That allows you to gauge the upper bound of enhancing something like resistance and place it so that the practical value of enhancing the power approaches a particular value no matter how much enhancement you manage to get. We now have a horizontal asymptote, ensuring that the composite function measuring the value of enhancements is concave down, and that returns are indeed diminishing.

Jesters Ghost wrote:

Well, using a bounded (expontential) model was suggested /waaaaaay/ back on the first page.
I don't see that that's the issue (or why it should still be an issue at least). At least within the domain of a single power.
Sure, resistance (and defence) are "perfect" when they hit 100%. So we use an exponential function (with an upper bound), and choose the base value of the power so that it can never hit 100%. Problem solved, surely?
Like I say, once you start combining powers, then you have the same issue all over again, but as Redlynne is correctly pointing out, that's no longer an issue with /how enhancements are attached to a power/. It's a fundamental problem with just how those properties scale.
It is /connected/, and a mistake in one place can exacerbate the other, which is why I feel is at least worth /acknowledging/ as something to be aware of, but it isn't an inherent problem with the original suggestion.

It's an inherent problem with the original suggestion [I]as written[/I]. All I'm saying in this regard is that the square root function would be problematic when used to apply enhancements to certain attributes such as resistance, and that a bounded function is needed. I know that was suggested as a possibility way back; I've been saying it's actually necessary to make that change, no matter what else you do, and providing an explanation as to why. I've also said that ED effectively took care of the same problem, while acknowledging that its other less-desirable qualities mean that a different solution is preferable. At the moment I think arctan is the best option because it is an increasing function with two horizontal asymptotes, which would actually allow you to get similar behaviour if you wanted debuffs to function as a negative enhancement. This is all I've wanted to adjust in this regard, though it seems as though people have interpreted that to mean I would scrap the whole thing.

Whether or not the problem still returns if you look at power stacking: maybe. If you know the upper bound on any particular resistance power then given a particular stacking mechanism you can predict the upper bound on the final result of a particular set of powers. If you allow powers to stack in a way that they approach 100% [I]and[/I] you cannot predict the set of powers that will stack then the problem does return.

I don't have time to do a detailed reading of this as I'm on my way out to work, so I will have to come back to it later, but I believe the problem is right here:

Redlynne wrote:

The Diminishing Returns formula that determines the "shape" of the diminishing returns "curve" when Enhancement "values" get stacked up (mandating that it be a curve and not a cliff)

This is the part that you said you've staked a position on that I believe is the source of the issue for me. In using a square root function, and in particular a function that is unbounded, it doesn't matter what power multiplier you use in terms of trying to get the behaviour you're after. The practical impact of resistance is such that applying an unbounded enhancement function on such a power results in returns that aren't diminishing; not in the sense that actually matters. A reasonable measure of the power's effectiveness based on number of enhancements will be concave up, and so enhancements don't really diminish in value as you add more. This behaviour is a consequence of applying a square root enhancement on an effect that would essentially have infinite value if you were hypothetically able to reach a particular total (in this case, 100%). The problem exists before any stacking issues are considered and regardless of what power multipliers you use.

Now, you [I]might[/I] manage to keep a lid on the issue when the power is taken in isolation, but I think it's essentially impossible in a practical setting (with power stacking, increasing number of global enhancements over time, and the like) as it's a moving target, and I think it's a mistake to introduce the problem in the first place. The crucial assumption to your proposal that makes it work is that your enhancement formula provides diminishing returns, but I don't believe it does in the way it interacts with certain attributes, such as resistance. A bounded curve doesn't have the same problem, which is why I believe it would need to replace or work alongside the square root function [I]within[/I] the context of enhancement calculations.

Look, we're obviously just going around in circles here. As far as I'm concerned, you keep "pointing out" the same thing as if I've failed to see it, and I keep saying I [I]know[/I] that, but it isn't doing what you think it's doing. While there is obviously an upper bound on the enhancement value in the sense that you will only be able to put so many enhancements into the power/global slots, it is the fundamental nature of the functions in question that leads to trouble.

I'm going to throw out particular numbers here just to illustrate the point, and not because the point actually relies on these numbers. Suppose you have a power (by itself) that provides 40% resistance with a power multiplier of 0.6. Suppose further that you have 100 hit points and we use the function I mentioned a couple of posts back to calculate the effective hit points that each enhancement provides. The results are as follows:

1st enhancement: 41.7 hit points.
2nd: 24.1
3rd: 22.6
4th: 22.8
5th: 23.7
6th: 25.2
7th: 27.2
8th: 29.8
9th: 32.9
10th: 36.8
11th: 41.6
12th: 47.5
13th: 55.0

By this very reasonable measure the returns on enhancements aren't diminishing, at least not past the 3rd enhancement as they start growing again. This is an inevitable consequence of the behaviour of the functions. In order to deal with it on a power-by-power basis you have to adjust the power multiplier not for the purpose of balancing the power around the efficacy that you want, but simply to try and enforce the diminishing returns behaviour that the system relies upon. In this particular case you would have to reduce the power multiplier a lot before you achieve this result. At 0.3 for example the bonus is very close to linear after the first few enhancements all the way up to 13, at which point they start growing again. You will routinely find that balancing the power to the approximate level you want is at odds with trying to enforce a diminishing returns behaviour.

The problem here is expecting the power multiplier to do the job of [I]both[/I] the enhancement schedules [I]and[/I] ED caps of CoX, and it simply can't with any level of fine control. This problem is already present before you consider stacking powers, but it is only made worse in that context. Depending on the way you decide to stack powers, you can easily find that a power for which you enforced diminishing returns up to a certain number of enhancements no longer behaves that way when stacked with others. Ultimately this is a problem with how the functions you're using interact with a property like resistance.

You could compare, for example, another power that directly provided a hit point buff. Suppose you had 100 hit points and are now enhancing a power that provides a 40% bonus to hit points with a 1 power multiplier. The number of bonus hit points by enhancement is as follows:

1st enhancement: 20.0 hit points
2nd: 8.3
3rd: 6.4
4th: 5.4
5th: 4.7
6th: 4.3
7th: 3.9
8th: 3.7
9th: 3.4
10th: 3.2
11th: 3.1
12th: 2.9
13th: 2.8

Here we see diminishing returns behaviour that will remain intact under any change you want to make to the power multiplier. The difference is a result of the fact that the underlying system being enhanced (and the way we value increases in that system) has a different behaviour.

Pyromantic wrote:

Suppose you have a power (by itself) that provides 40% resistance with a power multiplier of 0.6. Suppose further that you have 100 hit points and we use the function I mentioned a couple of posts back to calculate the effective hit points that each enhancement provides. The results are as follows:
[ ... omitted for brevity ... ]
By this very reasonable measure the returns on enhancements aren't diminishing, at least not past the 3rd enhancement as they start growing again. This is an inevitable consequence of the behaviour of the functions.

Sounds like your function is broken. This is not a fatal problem. What was the function again?

Pyromantic wrote:

The problem here is expecting the power multiplier to do the job of both the enhancement schedules and ED caps of CoX, and it simply can't with any level of fine control.

It can if your underlying function works properly. Generally, if you want your function to be adjustable at N points, it needs to have N factors that can be adjusted. Exponential curves have at least three adjustable factors: scaling, asymptote, and base. Scaling doesn't have to be linear.

Further, functions can be combined to provide ganged effects. It all depends on what points you want the curve to hit.

Pyromantic wrote:

This problem is already present before you consider stacking powers,

These equations aren't meant to also be used for stacking powers.

One equation for diminishing returns and scaling of a single power on a single person. Offensive or defensive.

Another equation for diminishing returns of stacking.

They don't have to be the same equation.

Pyromantic wrote:

Depending on the way you decide to stack powers, you can easily find that a power for which you enforced diminishing returns up to a certain number of enhancements no longer behaves that way when stacked with others.

Yes, a broken algorithm is broken, but that does not mean all other algorithms are also broken.

*Head 'esplodes*

Comicsluvr wrote:

*Head 'esplodes*

"Ah, yes, his head's been ripped off. I'll get you another."

(EDIT: changed link to actual sound file.)

Lin Chiao Feng wrote:

Pyromantic wrote:
the function that measures the value of resistance as a multiplier of effective hit points (1/(1-x)).
Well, there's your problem. That function is supposed to be 1 / (1 + x). So it gets smaller as x gets bigger, but never quite reaches zero. That's what you mean by "multiplier of effective hit points", right, that this number is then multiplied by the incoming hit points of damage to determine actual scored hit points? if not, you really need to clarify, maybe by showing all the math you're positing.
In any case, the function you have increases without bound toward infinity at x = 1, which is a red flag that the math is wrong.
And if you're taking Redlynne's boosted value and dropping it right in for x, you're either skipping an intervening soft/hard cap step or something similar. And you are definitely well beyond the scope of the original proposal.

I did explain that math in post 150. One way to measure the value of resistance is to look at it as a multiplier of your hit point total, which is the function I'm using. If, for example, you had 75% resistance, then that resistance provides you with effectively 1/(1-0.75)=4 times as many hit points.

The fact that there is a vertical asymptote at x=1 is exactly the [I]point[/I] and a good indication the math is [I]right[/I], because if it were possible to have 100% resistance then its value would essentially be infinite from a defensive standpoint; no amount of incoming damage would be sufficient to hurt the character.

Apply a hard or soft cap if you want, but all that would mean is that the value would just drop to 0 for enhancements beyond a certain point. Presumably you don't want that to happen inside the enhancement of a single power (note that this still has absolutely no power stacking involved). If you want to change this behaviour then you need to change the function or apply another one on top of it, which is what I said way earlier in the thread--that the diminishing returns behaviour should be happen within the enhancement system itself.

Redlynne wrote:

Your effective hit points multiplier is merely one way to go about judging effectiveness, but it is not the only way, nor is it even necessarily the best way (per se) of making this judgement. In the absence of a structure defining how Resistances "work" in respect to incoming damage ... because the rest of the entire combat system hasn't been defined yet ... you're basically writing on liquid water with your finger. And since we already have exploding heads, I'll spare everyone the effort involved in basically repeating what I've already written before.

Nope, not the only way or necessarily the best way, but this behaviour is a necessary consequence of any measure of resistance that considers 100% resistance to be infinitely valuable, which I think is entirely reasonable under anything resembling CoX's percentile resistance system. I'll repeat again here that it doesn't matter that you limit the number of enhancements so that you can't actually reach 100% resistance; the fact that it is hypothetically there anyway means that this eventually-growing-returns-on-enhancements behaviour will occur.

Ultimately, though, if you want to say that all of this depends on defining the way resistance works in the game, then all you've done is come around to the point I made in the [I]first[/I] place, which is that you can't really determine how enhancements should work until you figure out the underlying systems that you're enhancing.

Pyromantic wrote:

I did explain that math in post 150. One way to measure the value of resistance is to look at it as a multiplier of your hit point total, which is the function I'm using. If, for example, you had 75% resistance, then that resistance provides you with effectively 1/(1-0.75)=4 times as many hit points.
The fact that there is a vertical asymptote at x=1 is exactly the point and a good indication the math is right, because if it were possible to have 100% resistance then its value would essentially be infinite from a defensive standpoint; no amount of incoming damage would be sufficient to hurt the character.

So your claim is:
[list]
[*]Posit Redlynne's proposal.
[*]Assume an off-topic algorithm in which the proposal creates an absurd result.
[*]Therefore the proposal is bunk.
[/list]
All you have proven is that your assumed algorithm of resistance is unworkable, not that [i]all[/i] algorithms of resistance are unworkable. Maybe, with a cap, it resembles the CoH algorithm. That does not in any way mean that there are no other ways to do it that will work.

When you step back and look at it, that algorithm was one of the biggest problems in City of Heroes. If there weren't a cap around 50%, pretty soon each additional percent of resistance would have an overwhelming amount of damage mitigation. This was an [i]accelerating[/i] returns curve! And without the cap it would have done all those evil things you spelled out.

So what if we just don't use it? What if resistance weren't a percent of incoming damage absorbed, but say a percent bonus to the "effective number of hit points" (or however it should be phrased)? What if having "100%" resistance meant you had effectively a 100% bonus to hit points? You get percentages [i]and[/i] you get no implicit upper bound. 1/(1+x).

Boom, the bounds craziness problem goes away. Problem solved, rampant celebration, there is happiness and light in the land again.

This is approximately where Red would have gone if it weren't for the fact that [b]it is a digression completely off the original topic and will eventually lead to a discussion of all the game mechanics,[/b] none of which have even been determined much less publicized.

Pyromantic wrote:

Nope, not the only way or necessarily the best way, but this behaviour is a necessary consequence of any measure of resistance that considers 100% resistance to be infinitely valuable, which I think is entirely reasonable under anything resembling CoX's percentile resistance system.

Pyro, I have no idea why you must approach all of this in such an adversarial fashion. If the mechanics are broken, just fix them. There's no reason to assume the devs will use broken mechanics.

And this, THIS is why I won't do two shows a day any more. Nope...won't do it...

Your idea for Global Enhancement Slots looks eerily similar to a suggestion I posted on the old forums back in april. :)

Xander Cross wrote:

Character Boosts - Would be similar to enhancements, but for the character itself, not the individual powers of the character. A character would essentially have 5 character slots for boosts. They start the game with 1 slot available, and one slot unlocks every 10 levels thereafter. These boosts would enhance a character "stat" like increased health, lower recharge, more damage and so on.

Segev was in charge at that time, so things may have changed since then... but if it helps, this is what he had to say about it:

Segev wrote:

Character-wide boosts, as you suggest, are something that gets bandied about periodically. It's under consideration, though coding limitations and balance considerations may ultimately dictate whether it's a good idea or not. But we are thinking about it.

Overall yours looks like a good system on the initial pass (I haven't re-read it yet). I think they've already established what they're wanting boosts to do (already had a good idea of it back when I made my suggestion), and were planning on doing something that sounded like it might be quite different from CoH's implementation of enhancements. Though whether or not the plan changed, I've no idea.

The (then) studio manager, PlanetJ also had this to say about boosts... again, focus may or may not have changed.

planetJ wrote:

"Gear" in this game will [b]not[/b] do the following:

1. Affect appearance

2. Be outleveled...(it will always work, it just won't work as well as something newer...)

Also, we will not have any sort of "global stats" in the sense you think...you won't be putting "points" into any stats in this game.

Because the only "stats" will be from "boosts" which you can always tinker with...there would be no need for a system that self perpetuates. Boosts are going to be our version of "Gear" and I think you'll love the system we have cooked up for those. If you like to customize things...you'll be amazed at what we can do with our "Gear" and it won't be quite like anything you've ever seen before!

Redlynne wrote:

"Great Minds Think Alike."

Indeed, and fools seldom differ. ;)

" I think the one thing we can all agree on is that further discussion is pointless."

Well you're entitled to your opinion...

" However, I am using percentile resistance as a particular example to demonstrate that the proposal wouldn't behave the way it's designed to when interacting with some mechanics. That doesn't mean there's necessarily anything wrong with percentile resistance."

It also doesn't mean that the idea wouldn't work well with a number of OTHER mechanics not even discussed yet. It sounds to me like you're designing the bathroom before the house. What you're saying might be true IF the mechanics works the way you say they do. Of course if the mechanics are totally different then your statement is baseless.

"The issue with resistance is only a case in point selected because it is relatively simple to analyze..."

Apparently it isn't because we have pages and pages of theoretical mathematics and nothing to show for it because the system hasn't been BUILT YET.

I'm a Quality guy and that means I like things to be simple. ED sucked but it was simple, at least on the surface. There were tons of calculations running in the background of every action to determine what happened to who and by how much. The Blast does 50 damage base, with Damage slotted to +50% for 75 damage base. Of course we're using the AV code which automatically cuts incoming damage in half blah blah ad infinitum.

I understand that you guys are trying to get this right early because changing it later will suck. However nobody can say how effective anything will be until we have SOME sort of an inkling or a framework in place by which to measure things.

Comicsluvr wrote:

" I think the one thing we can all agree on is that further discussion is pointless."
Well you're entitled to your opinion...

Well, I said that because no matter how much I clarify my position it never seems to move the discussion forward. However, since I do prefer that my position be clear when any judgment on it is made, I feel the need to clarify just one more time. Just one. I hope. I probably should just give it up (and perhaps we all should), but it is what it is.

Comicsluvr wrote:

" However, I am using percentile resistance as a particular example to demonstrate that the proposal wouldn't behave the way it's designed to when interacting with some mechanics. That doesn't mean there's necessarily anything wrong with percentile resistance."
It also doesn't mean that the idea wouldn't work well with a number of OTHER mechanics not even discussed yet. It sounds to me like you're designing the bathroom before the house. What you're saying might be true IF the mechanics works the way you say they do. Of course if the mechanics are totally different then your statement is baseless.

I'm not sure what statement exactly you're referring to, but I do think I should acknowledge a need to adjust something I said earlier. I said at one point that it was necessary to change the function or introduce an additional one to something bounded "no matter what else you do," which isn't quite right. I think it's necessary in the sense that you need these kind of options. No one particular function is strictly necessary, and no set of functions is necessarily sufficient. The point is that different functions are needed for different tasks, and so you need a toolbox to make this system work. Just using square root (or something else) is insufficient, and the introduction of the power multiplier doesn't change this fact. It essentially allows you to change the magnitude of the curve in question, but not the shape of it. It is analogous to saying that my round peg is ok for all the square wholes because I can change its size. Yeah you can shrink it down to fit, but it still isn't the shape you want. From that point of view, I consider something bounded like arctan a necessary addition because it seems very likely that some mechanics will require bounded enhancement.

That's the point I've been getting at with resistance, not because it is [I]necessarily[/I] the way resistance will be done in the game, but because [I]if[/I] you were to do it this way then the system as it stands doesn't meet its stated goals. So I will say that I can understand why it appears that I insist resistance work a certain way since that's what I've been using to illustrate my point, but it has never been my intention to suggest it's the only option. Far from it, but percentile resistance is as good a place to start as any because that's the way it was done in CoX. No, that obviously doesn't mean it has to be done that way moving forward. It also doesn't mean that the 1/(1+x) function is the way to go. Could be that resistance is a value subtracted from incoming damage instead of multiplying it by a fraction (like how DR works in D&D) or something else entirely. And in each case, the needs of the system would dictate a different approach to enhancement.

This, as I've been saying, is because the way the math measures changing values is categorically different in each case. Even limiting ourselves to increasing functions, they might be concave up, concave down, linear, step functions, or something else entirely (concave up in some intervals, down in others; and so on). We don't measure the returns on enhancement by the percentage applied to the base value; we measure them in terms of the practical change in value of what we're enhancing. In essence, under this system, the real value of an enhancement can only be measured using a composition of functions involving the function determining enhancement [I]as well as[/I] a function measuring the value of change in the relevant property, be it resistance or something else. As such, the behaviour of the game's systems is inextricably linked to enhancement, so you either need to settle those first or start building a toolbox of enhancement functions, while also recognizing you might need additional tools down the line.

Comicsluvr wrote:

"The issue with resistance is only a case in point selected because it is relatively simple to analyze..."
Apparently it isn't because we have pages and pages of theoretical mathematics and nothing to show for it because the system hasn't been BUILT YET.

My use of the word "relatively" was deliberate, because it's still a substantial endeavour to analyze, but I think it's easier than, say, CoX's defense system, for a couple of reasons I can think of. (Firstly, that you are dealing more directly with probabilities, and measuring the value of probabilities is immediately trickier; and secondly, I don't think you can avoid introducing accuracy and tohit buffs into the mix.)

However, when you say the system hasn't been built yet, and further...

Comicsluvr wrote:

I understand that you guys are trying to get this right early because changing it later will suck. However nobody can say how effective anything will be until we have SOME sort of an inkling or a framework in place by which to measure things.

...you're actually echoing my point, made in the first post I made in this thread, which said this:

Pyromantic wrote:

To some extent this looks like putting the cart before the horse, since I don't know how concretely you can design enhancements until you know in more detail what you're enhancing.

Lin Chiao Feng wrote:

Comicsluvr wrote:
*Head 'esplodes*

"Ah, yes, his head's been ripped off. I'll get you another."
(EDIT: changed link to actual sound file.)

I get an amazon shopping page with no information about anything ...

Ebon_Justice wrote:

I get an amazon shopping page with no information about anything ...

My hat's off to the clever folks at that site. They randomized part of the URL, so it expired. Further, it redirects any old links with a Referer header to an Amazon page. (You get a "your link is dead" page if you just copy and paste the URL into the address bar.)

Sorry about that.

The site is www.wavsource.com, navigate to "Monty Python's Flying Circus", then about halfway down there's a link titled "Ripped off".