New Onslaught round scoring?

UT2004 related discussions
User avatar
Wormbo
Posts: 384
Joined: Sun 28. Aug 2011, 11:52
Description: Coding Dude

New Onslaught round scoring?

Post by Wormbo » Sat 19. Jul 2014, 06:59

I just saw an interesting post on the UT4 forums about a modified Onslaught round scoring mode:
Veggie_D wrote:ONS (and WAR) scoring has always bothered me a little too, and I've long wondered about how best to improve it. Here's an idea I've been kicking around for years:


Point values:

3 for regulation win
2 for overtime win
1 for overtime loss
0 for regulation loss


GoalScore (scoring limit) values:

2: Match always lasts exactly one round.
3: Match lasts one, two, or three rounds. Possible final scores: (3-0, 4-2, 5-1, 5-4).
4: Match lasts two or three rounds. Possible final scores: (6-0, 5-1, 4-2, 4-1*, 6-3, 5-4).
5: Match lasts two or three rounds. Possible final scores: (6-0, 5-1, 5-3*, 7-2, 6-3, 5-4).
___
* Result of a final-round draw (see below)


Tiebreaker option:

When a match is tied at a score that is 0, 1, or 2 points less than the GoalScore, such that the next round will be decisive, the server may optionally begin the tiebreaker round in overtime, or with a very short (1-2 minutes) regulation period, to hasten the outcome and ensure that a match is not too prolonged.


Final-round draw:

If team A is within one point of the GoalScore and leads team B by two points or more, then team B must win the next round in regulation, or lose the match. For example:
  • GoalScore=4, team A leads 3-0 after one round: even if team B were to win round two in overtime they would lose the match 4-2, so they must win the round in regulation to tie 3-3 and force a third round tiebreaker.
  • GoalScore=5, team A leads 4-2 after two rounds: even if team B were to win round three in overtime they would lose the match 5-4, so they must win the round in regulation to win the match 5-4.
In those cases, if the clock hits "00:00" then the match is over: each team is awarded one point, which causes the leading team to reach the GoalScore and win the match.



What I like about this format, especially when you choose a GoalScore of 5 or 4, is you're always guaranteed a controlled, predictable match length — except in the case of an overwhelming victory, in which case it's just better to end it quickly anyway.

I'm a fan of the fast-tiebreak option as well, as I would find the strategy adjustments really interesting going into that final round. A huge emphasis would be on speed, efficiency and teamwork right from the beginning, which I think would make for some exciting finishes.
Onslaught and Warfare communities are in an excellent place to actually try that out already, as it should only require a modified Onslaught/Warfare game mode based on all the existing infrastructure. Do you think it'd be worth trying this with a customized Onslaught mode? (I'm cross-posting this to Omnip)o(tents forums.)

User avatar
EmanReleipS
Administrator
Posts: 2371
Joined: Wed 2. Oct 2013, 23:18
Description: Pancake Fairy
Location: Germany

Re: New Onslaught round scoring?

Post by EmanReleipS » Thu 31. Jul 2014, 14:37

I had a reply formulated in my head weeks ago, but of course I forgot to post it. :crazy:

My first point of criticism about this is that it's rather complicated. I've read through it several times now to make sure I got it right, but in general, I think it's best to keep rules short and simple. This seems rather complex to explain to a newbie who asks about the scoring mode on the server....

The second one is that I still don't see how this is an actual improvement over the original "2 points for regular time win, 1 point for overtime win" scoring mode. It seems to me exactly like the regular scoring mode, only with a few more points and a system to shorten matches if the score difference is too high.

Now, you mentioned that this would lead to a more predictable match duration, right? The match ending after regular time when one team leads by 2 points or more does indeed shorten things down and lead to a predictable length for the final round. But it doesn't help to predict the overall duration of all rounds combined, because with a goal score of 3, 4 or 5, you still get several rounds that can go into overtime, and if one team doesn't have a big score lead on the other, this "final round draw" won't happen (if I understood it right).

I'm sorry, but I don't quite see the merits of this scoring mode. :(
Image

pull0$
Posts: 23
Joined: Sat 26. Jul 2014, 22:17

Re: New Onslaught round scoring?

Post by pull0$ » Thu 31. Jul 2014, 16:53

EmanReleipS wrote: It seems to me exactly like the regular scoring mode, only with a few more points and a system to shorten matches if the score difference is too high. (
Actually it is greatly improved, because it rewards the effort of a) the losing team for not letting the offenders get to the core, b) the winning team for pressing hard in order to win 2 extra points - 3 in total - instead of 1.
Not particularly useful in single round mode ofc, which IS the proper mode in CEONSS for my tastes...

User avatar
EmanReleipS
Administrator
Posts: 2371
Joined: Wed 2. Oct 2013, 23:18
Description: Pancake Fairy
Location: Germany

Re: New Onslaught round scoring?

Post by EmanReleipS » Thu 31. Jul 2014, 17:24

pull0$ wrote:
EmanReleipS wrote: It seems to me exactly like the regular scoring mode, only with a few more points and a system to shorten matches if the score difference is too high. (
Actually it is greatly improved, because it rewards the effort of a) the losing team for not letting the offenders get to the core, b) the winning team for pressing hard in order to win 2 extra points - 3 in total - instead of 1.
Not particularly useful in single round mode ofc, which IS the proper mode in CEONSS for my tastes...
Yes, and at first I thought that would be nice. But then again, the score difference is still the same. Whether the winning team gets 2 points and the losing team gets 1 is still kinda the same as the winning team getting 1 and the losing team 0. The only difference being that it rewards the losing team for making it hard on the winning team and only losing in overtime. That might be more motivating. But that's it. We already have a reward for winning in regular time in the regular scoring mode (2 points instead of 1).

EDIT: Actually, it does make a big change when it comes to one team winning in regular time and getting 3 points vs. the losing team getting nil. But doesn't that big score difference make it really tough on the losing team, esp with the "final round draw" ominously looming on the horizon, thus giving them less time to win?
Image

User avatar
Wormbo
Posts: 384
Joined: Sun 28. Aug 2011, 11:52
Description: Coding Dude

Re: New Onslaught round scoring?

Post by Wormbo » Fri 1. Aug 2014, 16:19

That final round draw really only shortens the game without actually taking any chances away from any team.
Consider score limit 4. Team A scores in regular time, then forces team B into overtime and thus wins the match 4:2. What if the over time happened in the first round and team B won it? The score would be 2:1 and if team A scores its regular time win now, the game would have the same outcome - team A wins the match, this time with 4:3.

Or let's take goal score 2 of the classic scoring mode. Right from the start, each team may lose the match by losing a round in regular time. If you can't win, you can at least try to hold up a stalemate on your prime to force the game into overtime and get a second chance.

Try to look at it not just from the 32 player perspective, but also from a competitive point of view, where team sizes are much smaller. Competitive Onslaught could be played on slightly smaller maps and with shorter round times. It really pays off then to have a "best of 3"-ish scoring mode like goal score 4 or 5 with the proposed system provides.

User avatar
Pegasus
Posts: 1148
Joined: Wed 4. Nov 2009, 23:37
Description: ONSWordFactory
Location: Greece

Re: New Onslaught round scoring?

Post by Pegasus » Sun 3. Aug 2014, 15:35

I know it's taken me a bit longer to comment here than it usually does as far as threads dealing with the more theoretical side of ONS go, but I gotta admit that scoring systems in tournaments/sports isn't exactly my forte, or even a field I've had the chance to devote much thought on in the past in general. That's why when my first reaction upon reading this proposal was a blank, glazed-over gaze, I knew it'd probably be best to ruminate some more on it and maybe examine others' readily available reactions and analyses to this from all the related/cross-linked threads than to just rush in and proceed to put my foot in my mouth by blurting the first thing that crossed my mind about it. Even though discussion of this idea didn't spread too far, after awhile I did manage to settle on a few thoughts and critiques about it by tying the concept into some neighbouring ONS aspects I've had given much more consideration in the past; trouble is, by that point I was also well into vacation traveling preparations, so the whole thing got shoved in the proverbial back burner again until I could establish some kinda summertime routine and be able to find the time to start working on my ever-growing pile of pending "srs web bsns". Just managed to get started on that now, and I've already been told twice that I'm "doing vacationing wrong" for even considering opting out of certain customary summer activities in order to get some proper online writing done, but it's not like those folks can easily relate to fulfilling longstanding hobby responsibilities to a large group of people, so eh. Anyway, point of all that is, sorry for the response delay there, it wasn't out of indifference, so here's some, hopefully still useful, feedback on this now.

The main gist of this scoring modification idea seems to revolve around trying to ensure a maximum match length cap through awarding a point to the losing team during overtime, as well as employing some optional and potentially time-saving measures for matches' final rounds, but it's a system I'm reluctant to embrace for one big reason. See, while the goal of preventing matches from turning into arduous slogs beyond a 3rd of 4th round is indeed commendable, I cannot overlook the fact that the manner by which it attempts to inject the additional motivation for expedience to players is through an IMO problematic incentive-disincentive pairing. By this I mean that during any "match round" (that is, any round where the added points upon its conclusion would be enough to end the match) where the winning team leads by more than 1pt, they would no longer need to worry about getting out there and dominating the battlefield once again, but, instead, they'd know they can afford to hang back, turtle or mess about in any other way to the detriment of gameplay quality (I suspect little would corrode morale faster than a team-wide unsportsmanlike/insincere performance), and the match would still end in their favour as soon as the timer reached zero just by virtue of the scoring math. Obviously, this wouldn't necessarily always have to be the case, but the room for that is there. I realize this may come across to some as trying to deconstruct a seemingly technically sound model through moral[istic?] means, but I'm honestly concerned about there being a flaw in it that could jeopardize healthy gameplay in a proportionally not insignificant amount of matches down the road for any server adopting such a modified scoring system. When one team's incentive to win is countered by the other team's opposing, but equal motive, we can have the makings of healthy competition, but when teams know the numbers can afford them to half-ass it and still manage to grab first place in the end, well, anyone who's seen any Formula One races past a year's mid-point (or followed any major football league past the 3rd quarter of its seasonal span) could easily tell you how that can cause emotional/performance investment to plummet, and, frankly, I don't see why the same principle wouldn't apply in a virtual setting just as much - albeit, adjusted for scope and scale, of course.
Main objection aside, I also gotta agree with Ema here on this system being liable to confuse a lot of players for awhile before they'd manage to internalize all the scenario and sub-case combinations applicable to any ONS match they'd be in. I mean, just thinking about potentially exposing our ingame admins to a barrage of uninformed remarks about unexpectedly shortened tie-breaker rounds to the tune of "wtf, why's there no overtime, we were totally gonna win this, enough with CEONSS' custom code crap ruining the game!!11!" already makes me feel sorry for them :/. This would need some mightily visible warning messages ahead of time just to curb the majority of all the complaints, that much I can foresee for certain.


Specific potential drawbacks of the proposed scoring system's aside, I'd like to focus some more on the greater issue it aims to help with for a bit though: improving players' ONS experience through optimization of matches' duration. This, of course, is by no means a newly introduced concept and there's certainly been both a number of different (prevalent) theories across server management teams, as well as approaches employed in the hopes of delivering the best result. One common notion that most of those previous efforts seemed to had chosen to rely on, however, was that the answer should be expressed in the form of the ideal number of rounds from which players would derive the most value out of any map. Secondary to that would often also be the consideration of the "moral satisfaction" component, most commonly manifested as "room for a rematch", typically to satisfy the aggrieved (first round) losing side, or the need to confirm the dominance of one team over the other as not accidental/luck-based through the "best out of 3/5" practice.
While I can understand and appreciate the physical sports-based conventional wisdom/experience from which those practices were drawn and how they ended up informing most popular servers' configurations with regard to GameScore, regulation time and round scoring (say, Omni's 3pt matches and 10min RTs or DW's 1pt victories regardless of winning conditions), when deferring to my own, years-long impressions and personal observations of others playing alongside me, I could never manage to see how any of those theories successfully translated into getting closer to that original goal of matches ending with most people gratified on or above the average. If anything, the pattern of typical moods and reactions that I began noticing over time pointed to a different standard of tolerance and enjoyment, and one that, rather than observing the changes of rigid rounds' numbers, was simply based on time played.
More specifically, representing the impression I have of this as a simple graph would probably look like the same as a part of the infamous "uncanny valley" curve, only in this case, it would be starting from the smallest values on the horizontal axis and ending right in the middle of the valley's own trough; a steady increase in engagement can be expected to be observed in most matches as they progress, hitting a "sweet spot" maximum at some time point, and then to start dropping precipitously soon afterwards. What exact time point delivers optimal engagement and enjoyment for the average map I don't exactly know, and would likely need to employ statistical methods to ascertain (realtime logging, post-match surveying, large dataset crunching, etc.), but even empirically speaking, I'm convinced it's inside the window between 30 and 40-45 minutes. It usually will matter very little if that time point gets reached during the first [marathon] round or in the 4th short one, but as long as a match goes over 40mins, you can expect to see ppl starting to drop off, to mentally check out, to express disdain and consider themselves stuck in the same map, to see some individual players attempting to get everyone's mind on the next map by voting, to observe a portion of players, usually on the winning team's side, losing interest and start behaving less morally (camping, spawn killing, goofing around, picking verbal fights with opponents), as well as to start becoming much more acutely aware of any flaws and imbalances in the map. Of course I'm not suggesting here that matches lasting longer than 40-45mins will invariably turn into a lawless UT apocalypse or even that they'll be unlikely to been seen through to their conclusion on average, as there's a ton of day-to-day evidence that indicates otherwise; rather, it's much more frequent to observe the more subtle changes whereby, for many people, interest in how longer matches will end gets all but completely drained, and they'll simply be playing along mechanically in order to reach the post-match voting point, and I believe that we should be employing all the effort and tools at our disposal to prevent such server-wide mood drop scenarios, if possible and when already expected beforehand.
Based on that understanding, then, the goal, to my mind, is not to derive an ONS server configuration package (with any optional custom scoring system) that will ensure some maximum number of rounds or points, whatever those may be, but in picking any combination thereof that statistically delivers matches shorter than ~35mins, possibility of rematches and any suchlike criteria being secondary concerns beyond that. Despite these being thoughts that've been swirling in my mind, more or less fully formed, for years now, this is probably the first time I think I've gone into this topic at any serious length, internal talks included, so I can't help confess some delight that, even with the later addition of the EvenMatch mutator and the ongoing employment of standard ONS match length (20mins) here, CEONSS still manages to come out well under the 35min match time psychological barrier on average, even if scraping it on a few occasions. Maps rotating on average sooner than every half hour keeps things fresh, exposes players to ever-changing geometric, tactical and other kinds of challenges, and, most crucially, IMO prevents them from becoming too emotionally attached to a smaller number of specific maps that they would otherwise keep voting for on an endless loop if the repeat limit wasn't set to the value it's been. So even if we didn't deliberately arrange for it, hurray for small, unintended victories, I guess :).

To cut that rant off before it starts coming across as even more self-congratulatory, and return to the focal point of the thread though, while being acutely aware of the possible complications that such a departure from the traditional scoring system could bring forth would make me wary to recommend its adoption on CEONSS at this point, when it comes to evaluating its performance in terms of the metric laid out just above (as in, in terms of simple time span), I don't want to rush to any assumptions, simply because some matters just can't be responsibly approached by "eyeballing it". I think it'd be far more honest for me to admit that, in absence of it being realized and tested yet, there's no data to suggest it'd deliver shorter or longer matches on average, so in that sense, the only valid answer to that for now is, "I just don't know". Should it get made and start being tested somewhere though, you can bet I'll be keeping a keen eye on it and starting to crunch numbers the moment they'd start pouring in to see just how it performs and whether it proves to be an improvement on the classic approach.

Anyway, those are my thoughts on the subject of optimizing ONS matches' length; apologies for veering a bit off course to include some extra layers of theory, but at least I hope some of it helped a bit in context.
Image

User avatar
Wormbo
Posts: 384
Joined: Sun 28. Aug 2011, 11:52
Description: Coding Dude

Re: New Onslaught round scoring?

Post by Wormbo » Sun 3. Aug 2014, 20:38

Wow, the more complex the topic, the larger the wall of text, huh? ;)

Anyway, I really read it all and it's reasonable, as usual from you. Would you mind if I quoted you on the original UT4 forums thread? (Excluding first and last paragraph, as the audience there probably doesn't care too much about your well-deserved vacation.)

User avatar
Pegasus
Posts: 1148
Joined: Wed 4. Nov 2009, 23:37
Description: ONSWordFactory
Location: Greece

Re: New Onslaught round scoring?

Post by Pegasus » Sun 3. Aug 2014, 21:27

I demand the UT4 forums be made aware of my vacation, it's why I took it in the first place!

Nah, sure, go ahead, link/quote/share away if you feel any of it might help people's understanding of what makes for a better ONS match length :).
Image

User avatar
Wormbo
Posts: 384
Joined: Sun 28. Aug 2011, 11:52
Description: Coding Dude

Re: New Onslaught round scoring?

Post by Wormbo » Mon 4. Aug 2014, 21:01

Heh, made me chuckle:
GreatEmerald wrote:Hah, I just read that wall of text, and it was fun imagining him as someone with a top hat and a monocle, due to the writing style :D But yea, the tl;dr version would be "too confusing for new players; the real problem is match length, which should be around 35 minutes, else people get tired". Fair point!
Tea, Sir Pegasus?

Zon3r
Posts: 575
Joined: Thu 7. Apr 2011, 07:46
Description: Don't shoot at me!

Re: New Onslaught round scoring?

Post by Zon3r » Wed 6. Aug 2014, 18:14

The soring is good as it is, cause if on a really bad map(maps like lame a lot, loved only by ion campers) would be multiple matches, it would definitely kill the server, cause players would simply leave, i know i would
Image