Fundamental Change Discussion

[Jediluke]

So guys.

We, the admin, have made many fundamental decisions involving how the ladder works with out seeking the council of it's members. (that's what admin's do, especially before there are any members)

We originally decided to follow the IDL version when it comes to ranking.

1: Beat your opponent you take their rank

Option #2 was to follow Case's version of ranking which was:

2: Beat an opponent that out ranks you and move half the distance between your ranks. Which means only the 2nd ranked player can take #1 spot.


Please voice your opinion as to which you prefer and why or any amendments to the current version that would enhance the enjoyment and competition of the ladder.

(P.S. changes do not have to be implemented right away...if any are warranted. We can implement the changes on the following season. In fact, if any changes are implemented and it doesn't seem to be working out...we can always change it back or make tweaks for the next season until we have the single most perfect ladder ever created!)

[Jediluke]

My opinion.

I've participated on both of the above mentioned ladders and I liked both versions just fine.
I liked truly earning my way up the ranks of Case's as it was hard to do.
I liked the polar opposite of being able to take #1 spot in my first ever ladder match on IDL.

I'm ok with either version to be honest.
If we keep the current version I think I have an amendment that might somewhat combine both aspects in a way.

Minor change proposal:
Current ranking stays the same. Beat your opponent and take their position.

Change Proposed:
The only way to take the #1 spot is to beat the current #1 in a best of 3 contest. Any rank can challenge the #1 and take the #1 spot in a best of 3.

Level choices: Current #1 chooses 1 level. Opponent chooses 1 level. 3rd level is chosen by the Random Level Generator which is comprised of the top ten most popular played levels currently on the ladder (aka a neutral level)

This actually wouldn't be much of a change as the #1 ranked pilot almost always asks for an immediate rematch. It's special to take the #1 spot and it should be earned through a hard fought best of 3 scenario imo. If you get it that way, you really earned it in a definitive way. (no offense to those of us that earned it by playing one single match..which of course is all of us that have held the rank)

thoughts?

[DKH]

The ladder is fine how it is, I believe we are all having fun and enjoying the opportunity to play great pilots in a competitive situation. That being said, I do have a concern about the suggestion of best two/three matches for #1 rank. My concern is not because there will be best 2/3 matches played, but that a pilot will choose a level that would be considered unusual. I believe a best 2/3 scenario would be perfectly fine if BOTH PILOTS CAN AGREE on the first two maps played, then relying on random map generator to come up with third. If pilots want to be able to choose a level, then maybe we should have a list of levels that are acceptable for matches played on the DCL, like a pool of 50+ maps or something. That way if pilots choose a level it isn't some sort of cruel and unusual punishment. Considering the fact that I can build my own levels, why wouldn't I build something that only plays to my strengths? I don't because that would not be very sportsman like. Cheers!

[Drakona]

I don't like #1 always being a best of 3. It seems like a great way to guarantee whoever's at the top will have a hard time finding matches / finding time for matches, because who wants to always commit to 3? Plus, it comes with all sorts of logistical problems. What if you only play one? Are you stuck and can't play anyone else?

That seems contrary to our design philosophy -- like with vigorous defense -- of encouraging activity at the top.

My question is, what problem are you really trying to solve? That someone will take #1 in an angle level and then refuse to play anywhere else? They can't do that anyway. They have to respond to core D1 challenges -- using the randomizer if necessary, but they have to respond. And I mean, we don't have a rule yet for the timeframe they have to respond in -- kinda still getting a feel for the tempo this community functions at -- but I can't imagine everyone ultimately playing dumb while someone avoids matches.

Personally, I don't see losing #1 in unusual circumstance and having an unexpected pilot up there for a week as an actual problem. Someone will take them down in core D1, and then all will be well. And anyway, you don't have to take that kind of risk in the first place . . . that's up to what kind of a champion you are. I have a lot of respect for DKH playing LoNi_ in D2 Athena right after taking the top spot -- probably more than insisting on core D1 and keeping the spot would have garnered.

[LoNi_]

In my opinion we keep the way it is. I believe more matches would be played if we keep that way, and I guess the ladder will be a little stuck with that best of 3 game for #1 spot. I mean, the #1 won`t play more than those 3 matches in one day, and maybe we have 2 or 3 pilots trying to take a shot on #1 and won`t be able so fast, cuze will depend on when the #1 will be available again.

I vote to keep as the IDL style.

Maybe letting everybody see the algorithm rank will make DCL a new ladder where you can see the spot for the algorithm and for the IDL style.

[LotharBot]

I've got to side with the crowd here.

For someone in the #1 spot, being required to block out three times as much time in order to defend the title is going to discourage activity. We want to do the opposite, encouraging the #1 pilot to take all comers as quickly as they can.

For someone with legitimate hope of taking #1, being required to wait in line for the previous challenger to finish up to 3 matches means it could take considerably longer to get that shot. You might want to challenge tonight, but Djcjr beat you to it and that means Mark is going to be tied up all night and you might not see the winner of that series for another two days.

For someone who likes to play against top pilots but probably won't win against #1 any time soon, being required to play multiple grueling matches means I'm much more likely to seek out the #2 pilot instead. I don't want to have my whole evening devoted to getting whooped by the same pilot. And I think it could screw up both players' efficiency, as fatigue sets in once the second match seems comfortably out of reach.

(As for the rating algorithm, it's not really designed to determine who is in what spot, or predict whether LoNi_ or bahamut is better. It's only designed to tell the approximate strength of pilots relative to the crowd -- who's competitive with the top pilots, and who's competitive a level down.)

[DKH]

I hear everyone's thoughts, and personally I think the ladder is fine just how it is. Activity is key, and that's what I thought this was all about. Hopefully I can get on earlier to be more active with the top pilots. Cheers!

[RiTides]

I agree with keeping it as-is... I don't really like the best out of 3 idea.

Also, gold pilots are currently 69-0 against silver pilots!! (Since the few of us silver pilots who did scrape a win against gold, did so when we were gold before the rankings system was tweaked, or against other pilots who are now silver as well). Come on lads, we can do this... we need at least one silver to beat a gold before 100-0!

[Drakona]

I know, right? The skill ratings were designed to detect "who has a chance against who", but 69-0! C'mon, guys, surely the top of silver can eke out some wins against the bottom of gold. (And top of silver is . . . you, RiTides. Get on that, would you? )

<-- bottom of silver, can't help sorry.

Actually, silver's really competitive. I like it. Gold has a low end and a high end and a you-two-really-should-be-in-another-class end, but bottom silver seems to have pretty reasonable odds against top silver. That's pretty cool, though the skill classes are designed to be a bit wider than that. There's room in silver for pilots a fair bit worse than me (without dropping all the way to bronze), but we just don't have any. My theory is that I'm as bad as someone who has taken 1v1 seriously for a long time can possibly be.

Also of note -- Bronze currently has no wins against . . . anybody, really. Kinda makes me sad that they don't play each other -- there are medals, and a tournament, and season-end prizes and everything. And beyond wins against each other, I bet the top bronze pilots (spud, Blarget) could take down the bottom silver pilot (me) in the right place, on the right day. In fact, I played spud in a Rangers game yesterday, and it was a nailbiter!!



---------

Back on topic, about Cases vs. IDL style ranks, I feel pretty strongly it should be IDL style. I like the dynamism of the system. I like that people rise to the appropriate spot on the ladder very quickly. I mean, the rating algorithm doesn't know about ranks, and look how well sorted the ladder is! I like that it's well-understood, and I like that the higher-ranked pilot has something to lose every match.

Beyond that, I know I'm always saying we shouldn't do things just because IDL did them, and I do stand by that. But when it comes to very basic things like how matches are run and how ranks work . . . I think we should what IDL did, simply because it defined the community for so long. We should break away from that some, learn from the experience we had on that ladder, do things appropriate to the current times. But if we break away from it too much, it won't feel like the kind of Descent ladder we grew up on, you know?


---------

Back off topic again, about making the internals of the rating algorithm public . . . I'd really rather not do that. It's designed for estimating which pilots have a chance against each other, and it seems to be doing that job well. But it's not very fun to watch update in real time. It moves your scores slowly, and sometimes in surprising directions, and often due to circumstances beyond your control. You can have a GREAT match against someone, and your score doesn't change. Then they go off and lose badly to someone else, and your score drops.

I could be talked into publishing the raw scores every couple months, maybe, and I'd love to publish a technical description of the algorithm itself. But making it so everyone can watch the scores update in real time? I promise you, that's not going to be as satisfying as you think it is. Just ask Lotharbot or Jeds! They've been watching the scores update from the beginning to help me catch bugs and problems, and I've heard plenty of frustration about how their personal scores change match to match. Mine, too, really; unless you understand it as a broad estimate, the experience is disappointing and frustrating. You can't MAKE the numbers move by having a great day. You have to have a great season. The numbers are designed to track how you perform against the crowd, how you get better or worse, averaged over all your good and bad days, over all your good and bad matches, over the course of several months.

So yeah. It's doing a good job sorting the pilots into classes, but it isn't designed as an alternative rating system. I think if I published it, people would obsess over why they're slightly ahead of or behind someone else, and what changed it, and the answer is going to be unsatisfying: the algorithm isn't accurate down to the kill, and sometimes the math just comes out that way. Did it put you in the right skill class? Then it's doing its job.

I'd rather it stayed magic.

But I acknowledge that people are curious. Maybe I'll publish a snapshot? If you all promise not to obsess over small score differences too much?

[RiTides]