Hiyas!

I have a few questions about my understanding of decay and whether my misgivings about it are based on fact. I also have some questions about another solution I'm trying to figure out.

For a while now, our guild had been planning to use the automated decay system. We are almost to the point where peoples' points will start decaying, so I started looking in to it more to make sure I understood it, and found out I had no idea.

I now think I understand the system: all events decay, whether that event is a DKP credit (earned) or debit (spent). It seems that unless a player always spends their DKP the day they earn it, their earned will decay before their spent, putting them at a disadvantage until the spent points also decay. Is this right?

I'm not sure what the advantage is to having earned points decay before spent points, other than possibly punishing veterans more than newbies, which doesn't work well for our guild. Is there another advantage to this?

The biggest problems I think our guild will have with decay are:

• I can see it being extremely frustrating for players to be threatened with needing to spend their points "or else they'll decay," only to find out that after they are spent, they decay anyway, leaving the player to wait for the spent points to "catch up" before getting them back.
• I also think DKP should be very simple and very transparent - this decay system is very hard to explain even to the other officers.
• Finally, I'm afraid an individual's DKP will come and go in "waves." The player builds DKP over several weeks, sees it start to decay, and spends it all. Soon after, the old earnings go "over the falls" and the player's total is much lower, meaning they have to save. Suddenly the spendings go "over the falls," the player's total leaps back up, and they decide they'd better spend it quick. Repeat.

If I'm way off-base on any of the above please let me know!


Our guild does need something to control inflation, and I think the ideal solution for us is a system that only affects players who aren't spending enough after a certain time-frame.

I started thinking of ways to figure out how much DKP each player earned in the last 5 weeks, to manually tax anyone above that. I thought I could do it through a spreadsheed and/or custom DKP columns, and I created this custom column:

((earnedperweek * 5) - earned) + spent

Basically we start with your 5-week earned, take away your total earnings to get your out-of-5-week earnings, then add your spent to see if you've spent all the out-of-5-week points. If the result is positive, you're ok. If the result is negative, you get taxed.

The problem is that it doesn't include adjustments, which is how waitlisted members are getting their DKP (and I think how I'm going to have to add the taxes in). This means the tax cut-off will get wackier and wackier the more adjustments a person has, in favor of the player or not depending on whether or not I include total adjustments in the calculation.

The next best thing is to do all the calculations on a spreadsheet, comparing adjustment amounts from a 35-day Age Set to the total Adjustments, but when I copy data from the 35-day Age Set it only includes players that have activity in that range, meaning the player data from the two sets doesn't match up. =(

Is there any solution for this? Or any chance that some new custom column variables could be added for adjustments inside/outside 5 weeks? Thanks for your time!

My suggestion is to put a DKP cap in place (the system can automate that). And then only consider a decay system for people who have not been raiding for a set period of time. If you don't like the current decay calculations other methods of decay are on Chops white board for sometime this month.

--

I thought about a cap, since it's the simplest solution, but the people it affects most are the most active players, and I would really like a system that isn't biased against them. It affects semi-active players second, and doesn't affect inactives at all. The idea of pairing it with a decay for inactives isn't terrible, but is still far from ideal since it gets more complicated to manage, and the semi-actives get the best deal.

With regards to the current decay calculations, I'm not sure if I like them. After reading all 40-or-so posts that come up in a forum search for "decay", I'm still not sure if I understand it, though I'll never claim to be the brightest crayola in the box. =P I'm honestly hoping that by voicing my concerns someone will tell me they make no sense and why. =)

Between the whiteboard and this post:
http://www.dkpsystem.com/viewthread.php?threadid=8261&p=1#p23290
...it looks like the three new types are Linear, Percent, and Tax.

If I understand the current decay at all, Linear might reduce some of the problems with earned points decaying before spent, but wouldn't eliminate them. Tax looks like it still affects everyone (albiet more fairly and without wild fluctuations, people going negative, etc), unless there are options to tax only points left unspent after 5 weeks as I'm trying to do above. And Percent I won't pretend to understand at all, other than that it looks like another decay variant. =P

The added variables would solve all my problems and allow me to do this manually without a huge headache. Chops you're doing a great job, I know you have a lot on your plate, and I hopehopehopehopehope that a solution to my problem is hidden somewhere in those whiteboard scribbles. =)

Sorry for taking so long to respond. Between development and support questions, I had put this "big analytical post" in the back of my mind to deal with it when I had a little more than a few free moments. The decay discussions are always pretty big

In any case, you present the analysis from a different perspective. You're right, the earned will almost certainly decay faster than the spent, and the decay rate from day-to-day (and week-to-week) will be variable (sometimes decaying positively after a huge portion of points get spent). But this effect is reduced a bit when you employ the use of a "Full Period" (allowing each point earned/spent to remain at full value for X days). This gives your users time to spent their points before they get hit with decay. When users are getting decayed, the points that are decaying are old points that are decaying, and not the recently earned points.

Ultimately, the key philosophy into the decay system in it's current implementation is this: "recent events are more important than past events." If you disagree with that statement, that's fair, and the decay in it's current form might not be for you and your guild. But it's important to remember that any system of decay will almost always affect the veterans more than the newer members, since the veterans in most systems will almost always have more DKP to decay. A tax-based one (say 10% per week), will take off more points from someone with 80 points than someone with 8 points. It's just a little easier to comprehend "10% of your total per week" than it is to comprehend "an accelerating rate of decay on each individual earned and spent point, each decaying independently of the other" (there's a lot to understand in that sentence). But it is absolutely transparent (all data and information is completely revealed), it's just a little hard to comprehend. (particularly, when you're trying to think in aggregate).

To understand our decay system, it's easier to understand when you look at a single event (a single loot receipt, or a single raid attendance), and see how the points drop on that one attendance, independent of every other attendance or loot. Then simply realize that you've got a bagful of events that are decaying based on their initial date, and each is decaying independently.

Decay does have the side effect of controlling inflation by constantly working to reduce the points of everyone, affecting more severly those who've earned more than they've spent. Conversely, those who've spent more than they've earned will actually benefit a bit from decay (and that's a key side-effect: encourage users to spend to prevent hoarding)

It's important to realize that the only way to completely eliminate the effects of decay (where points are reduced more for those who raid more) in any system of decay is for each player to end with a zero balance at the end of each raid. There's no system of decay out there (that I can think of), with the exception of the system "decay only those who are inactive" that doesn't penalize more heavily those players with more points. Our current system of decay accomplishes both with a single formula - it keeps players attending, because if they don't attend at least some raids, they won't be able to hold onto their points for terribly long, and it also keeps players from hoarding points, because the longer they go without spending, the harder it will be for them to keep up with the decay when they start decaying hard (without the spent items decaying in reverse to offset the daily change).

On the other hand, if your primary concern is simply to reduce inflation, the "tax" based method that's coming up is probably the best approach for you. This will simply look at someone's current DKP, and reduce it by x% every y days. It'll do this for all players and all at x%, making it very easy to understand, and having the nice side-effect of not changing the ordering of players (which is one side-effect of the current aggregate decay rates, which in aggregate, become quite variable).

My guess is the DKP tax will become the more popular approach to decay on this system primarily because it's regular, understandable, and doesn't mess with the ordering, while at the same time, accomplishing the key goal of limiting inflation.

--
It's all in the reflexes.

I think I would prefer this type of decay.

This has been requested in the past, and it's something with with the coming changes to decay might actually be pretty doable.

This would be very similar to the "Tax" decay, and it would, upon further thought, be mostly trivial to add a conditional "Only if the user hasn't raided in X days" to it.

--
It's all in the reflexes.

"decay only those who are inactive"

It's what I do with my guild manually every week. Every weekend I check who hasn't raided in the last 30 days and then do a mass adjustment to remove 10 DKP from each of them.

At the same time we have a DKP caps (different caps for different ranks) to avoid hoarding.

After each raid I have to manually adjust all the ranks to their caps aside from the largest cap which DKPsystem takes care of for me.

It's a lot of manual work, but the theory works for us, so I do it.

In case anyone is wondering, the reasoning behind multiple DKP caps is to ensure that players that attend more get rewarded vs those that are either applicants to the guild or who rarely raid.

--

Chops - No problem. I appreciate that you even take the time, since I know I can be pretty wordy. Most companies only dream of having customer support that responds in the time-frame you manage on your own. =)

You're right with regards to those who spend more than they receive getting a benefit, and I think that may be why it doesn't fit for us. We auction gear for DKP to determine cost, with 0 being the floor, so no one is allowed to spend more than they have received. I think this is why the possibility of going negative, or being disadvantaged temporarily until your spent points catch up, is so jarring to our current DKP system.

Of course you're also right that most any system affects veterans more, because the idea of any decay is to allow noobs the chance to catch up. So long as our members are spending, noobs can catch up, so I think that's why I'm trying so hard to figure out a way to only affect those that aren't spending, and affect them according to their own attendance.

Any thoughts on exposing some variables in the custom columns interface, so that we can add adjustments in to the equation? =D