|
|
Rank: The Illuminist

Groups: Administration
Joined: 5/8/2006 Posts: 8,410 Points: 2,875 Location: Seattle
|
Reference: http://www.wdfnews.com/f...aspx?g=posts&t=10171The plan would be to remove the cap on points loss. We can't really remove it, but we can max it out (haven't checked what that is). This would mean that high players would lose massive points to newer players. We'd also go back to 1000 min kills if this was enacted to prevent "noobs" as Sunfox calls them from taking points from legit regulars. I like the idea.
|
|
|
Rank: Administration

Groups: Administration, Newsie, Registered User, Server Admin
Joined: 11/6/2007 Posts: 987 Points: 783
|
I'm reproducing the "Endyss Plan" from the thread Whut linked to here, so you can see exactly what he's talking about: Endyss wrote:I've noticed something about the way the point system is going -- it creates a nearly impossible hurdle for people to climb past a certain point. Though I don't know if HLstatsX can even implement it, some kind of point decay whereby Players with scores in the multiple hundreds of thousands don't just need to log in once a month and retain an almost insurmountable lead over anyone else would be good. I say this as someone who doesn't give a flying crap about his ranking, but just observes that it would be very hard for any normal player to reach three hundred and eighty thousand points, EVER much less by the time TF3 comes out. You can see what I'm talking about by looking at this graph of our top 300 players' points/rankings (as of today):  Obviously, this tells us that our core base of players is around 80 or so people, but it also illustrates the insanity of the upper bracket, for whom a 50 point loss becomes no real impediment. By Rank 7, you're already at 23% less than the top score, and Rank 17 is less than HALF of Rank 1!! That means anyone already in the tail end of the Top 20 would need to gain between 120,000 and 200,000 points just to catch up. Someone just starting to play on our servers today would easily be in the Top 50, almost into the Top 20, for the same work.
The easiest way to fix it would be to remove entirely the cap on point loss, but that reintroduces behavior shaping problems we talked about to death way back in the day. Someone like Ducktor BBQ would easily lose THOUSANDS of points in a single death to someone just passing the minimum kill #, meaning he'd either cut back on playing, dramatically alter the way he plays so he dies less, and likely be very frustrated in the process (you thought you were ticked when you lost just 50 points!). That said, I think it's completely possible that some players could (and looking at the rankings, definitely are) creating a monstrous point buffer that no number of 50-point-loss deaths can ever substantially diminish.Like I said, in the end it doesn't bother me a great deal anymore, and if people are generally happy with the way things are, then that's great. But I think that back in the day, if we had had no cap on point loss, overall scores would be much lower and some people would experience frustration at losing a bunch of points, but the ranking would be a fluid expression of both time played and skill.Ed The Wolf actually got a congratulatory news post when he topped Daniel for the first place slot. But I worry that we are quickly approaching a point where no one will ever be able to practically take that spot from someone, and certainly someone below the Top 20 would have no hope whatsoever. HolyJaw wrote:In the end, everything Endyss said.
|
|
|
Rank: Coathanger Abortion

Groups: {pDs} Member, Server Admin
Joined: 5/17/2006 Posts: 3,487 Points: 1,990 Location: Undisclosed Location
|
I don't give a fuck. It will be awesome to lose 10,000 points because some noob stickybombed me. Razor wrote:Killing people is good therapy.  
|
|
|
Rank: Administration

Groups: Administration, Registered User, Server Admin
Joined: 6/7/2006 Posts: 188 Points: 189 Location: Canada
|
Something important is missing from that chart. "Time played".
Ranks 20-1 have an average play time around 18 days.
Ranks 40-21 have an average around 10 days.
Ranks 60-41 have an average around 8 days.
Ranks 80-61 have an average around 5 to 6 days.
Ranks 100-81 have an average of 3 to 4 days.
Why should someone with 3-4 days play time have anywhere near the accumilated points of someone with 18 or even 30 days play time? The points are more about length of playing rather than any sort of skill. So trying to give newer players with less time more points... well, I don't know, especially since those points are almost solely going to be coming from your core 80 players who spend the most time on the servers.
I don't like the idea of points resetting monthly, but I *would* support a reset of them right now so that everyone starts on a level playing field. Maybe a yearly or bi-annual reset.
|
|
|
Rank: Polio Power!

Groups: Registered User, Server Admin
Joined: 6/14/2008 Posts: 666 Points: 1,416 Location: San Diego
|
what about ressetting on a quaterly schedule with some form of awards given? That would provide a bit better incentive than "if you want to play a game as a non-good class you're going to be bleeding points out of the anus."
Some of the folks don't need their ranks to ever be doubted, this is the top ten class. But at the same time, Doc BBQ hasn't been playing THAT long of a time. What he did though was consistently play on these servers and get damn good doing it. His rank SHOULD be higher than someone who's played for a few days and gets lucky taking out a few of the higher up players, and then just does so-so and gets in the top 20's.
|
|
|
Rank: You're gonna stick that where?

Groups: Registered User, Server Admin, {pDs} Member
Joined: 6/24/2008 Posts: 419 Points: 649 Location: Eden Prairie, Minnesota
|
I think you just want a pds/hax/gwdr person in the top 2... YOU'RE ALL CONSPIRING AGAINST ME!!!!! lol :D McCheese wrote:I've always said all women are 2 drinks away from full on lesbian action Lead Salad wrote:I never figured whut for a fairy..wait, yes, yes I did. Snipme wrote:erk cant have me my penis will annilihate his ass hole
|
|
|
Rank: Administration

Groups: {pDs} Member, Server Admin, WDFNews.com Official Mapper
Joined: 5/31/2006 Posts: 3,454 Points: 445 Location: Merryland
|
i say yes, this will prove who is the best player, not who plays the most.  Cantrip wrote:When I die, I want to be reincarnated as a woman's jogging bra.
|
|
|
Rank: Fecal Impaction For the Win!

Groups: {pDs} Member, Registered User, Server Admin
Joined: 11/27/2007 Posts: 1,625 Points: 2,067 Location: Birmingham, Alabama
|
Dragon wrote:i say yes, this will prove who is the best player, not who plays the most.
Your gf said she would not be for changing it so my advice is that you retract your statement if you know what is good for you.
|
|
|
Rank: Administration

Groups: Administration, Newsie, Registered User, Server Admin
Joined: 11/6/2007 Posts: 987 Points: 783
|
Sunfox wrote:Something important is missing from that chart. "Time played". Ask and you shall receive:  I narrowed the range to the Top 150, since as I said I really think we start seeing our "regular" players at about Rank 80, and adding in the Time Played data bears that out. Time played scales very well with the score until about 80 and then you start seeing massive variance. This is where skill comes in. Some players obviously play a TON and don't collect many points, while others play significantly less and stay up there (including the players at the very top). Unfortunately, the only way to actually account for this and make sure a relatively new person doesn't clean house on a regular is to raise the minimum number of kills to a level that lets in all the regulars while excluding most "new" players. I'm not posting another graph right now, but just eyeballing the data, it'd take a minimum kill number of 2000-3000, and considering Whut wanted to lower it to be more accommodating, I think this might be unreasonable. But any other way and players will be losing a ton to a stickybomb from a guy with 501 kills. HolyJaw wrote:In the end, everything Endyss said.
|
|
|
Rank: Administration

Groups: Administration, Newsie, Registered User, Server Admin
Joined: 11/6/2007 Posts: 987 Points: 783
|
As far as which is more important, Skill or Time Played, I think philosophically we need to decide what the Ranking system is designed for. If the point is to keep people coming back and playing as a means of growing the community, then Time Played should be the emphasis and the graph in the previous post proves the current system is more or less doing its job. If the point is to reward skill and promote regulars that do well, then we need to up the point loss and follow it with a reversal of the minimum kill # to keep skilled one-timers from decimating the regular players. HolyJaw wrote:In the end, everything Endyss said.
|
|
|
Rank: Administration

Groups: Administration, Registered User, Server Admin
Joined: 6/7/2006 Posts: 188 Points: 189 Location: Canada
|
See, this is the problem with hlstatsx. It tracks more than enough statistics that it COULD be designed to be completely fair on skill vs time played vs friendly to new players. And it does do a bit of this to some extent.
The problem is, those calculations ONLY mean something if the two players are within a very close point range of each other. But the system puts so much emphasis on total points, that the calculations go severely out of whack if someone has 10X the number of points as someone else. For example, a 5,000 point player killing a 300,000 point player would probably be getting 10,000 points for that kill based on the current system, but the system maxes out at 50. But then that 300,000 point player killing the 5,000 point player would probably end up with 1/100th of a point, but the system rounds up to 1.
A skill-based system would have to eventually penalize a 5,000 point player dominating a 300,000 point player, and reward a 300,000 point player for finally killing the 5,000 point player. But that can't happen with the current system. The points awarded are meant to be based on "skill", but they're being calculated based on "time played". Which rewards new players (not necessarily bad), and ticks off regular players (not necessarily good).
I have an idea for a new system using the same data that is currently collected by hlstatsx, but don't know anyone who could implement it.
|
|
|
Rank: You're gonna stick that where?

Groups: Registered User, Server Admin
Joined: 7/3/2008 Posts: 190 Points: 303 Location: Pennsylvania
|
The discrepancy in points within the Top 50 is pretty large already, not taking into account the other 600 making up the riffraff of the 1000 kills league. As of now, if Rank 1 is killed by Rank 50, Rank 1 loses a minimum of 19 points, but with any normal melee weapon, taunt, or reflected shot, loses the cap of 50 as it is now. Things should be weighted as to how important they are to the game itself. Huge point values might work in a deathmatch setting, but the goal of Team Fortress 2 is typically some form of attack/defense territory control (except for you 2fort bastards where it is a deathmatch). Capturing a control point and helping your team reach its objective is worth 50 points whilst Rank 1 getting killed by a reflected rocket of Rank 650 is worth -1175 points. Is it really balanced in terms of the game's objectives that killing one person is worth the same number of points as capturing a flag or control point 24 times? I don't think so. Even now, at rank 8, I have trouble breaking even in games of 12 players or more, I don't think going from +10 or -10 points per hour in large games to -4500 points per hour will be much fun. This will just promote more turtling or objective-ignoring from most players which will make games worse. It's hard enough to coordinate any sort of team movement towards the objective as it is now. If everyone were to stay in the spawn the entire round out of fear of losing 1000 points, nothing will ever happen. Some servers have a point cap on the points that can be lost by the killed but not on the points gained by the killer. For instance, if I killed someone with a much greater number of points, I would get, say, 200 points for killing him or her, but that person would only lose 50. I'm not a huge fan of this method because it inflates scores too much, but at least it will ultimately balance them more if that's the plan. Ultimately, though, I'd be most happy with raising the kill threshold back to 1000 kills and keeping scoring the way it is. If the point cap is removed, though, think about significantly raising the point values for in-game events like kill assists, captures, et cetera. This might not keep scores lower, but it will keep them more balanced. Now, we 'high rankers' get the vast majority of our points from objectives like asssists and point captures while 'low rankers' get a greater percentage of their points from kills as you go down the ranks. The risk-reward for trying to play the game as it is meant to be played for most middle/high ranked players will be too high for some players.
|
|
|
Rank: Fecal Impaction For the Win!

Groups: Registered User, Server Admin
Joined: 1/13/2008 Posts: 1,509 Points: 2,004
|
I voted Yes. With every month the top 20 players just lock in and entrench the scores. I have noticed that the top 20 players also we tend to form our own games ...not ostensibly to protect our scores...but its a secondary benefit that we all have fun playing together...and we dont have to worry about being point farmed on Temp2 , or whatever. I don't really care about my rank....as it is now I will pretty much stay where I am at unless somebody gives up playing.....there are 20k-70k spreads between top ranks....Fuck trying to bridge that gap. Quote:
"Who the fuck is Leon Switch and why does he know we have a dog?" - Mrs. Giller
|
|
|
Rank: Administration

Groups: Administration, Registered User, Server Admin
Joined: 6/7/2006 Posts: 188 Points: 189 Location: Canada
|
Yeah, I already hesitated to play on 2fort because there was NO way to make back lost points if there were too many recent "normal points" players on the other team, but it's going to be 10x worse now. As much as the points mean NOTHING it still feels BAD to see them go down. Even holding steady is fine by me.
Let's say one of us top 20 players has 100 outright kills in a night. And let's say we have a 1.5:1 KD ratio. That would mean 75 deaths. Let's say 50% of those deaths are by people collecting 50 points (which is quickly going to be a reality during prime time). That's about ~1850 lost points in a night. Now let's see how we can make those points back:
1) Kill 1,850 people individually. As few as 925 people if we get the bonus points. 2) Kill Ducktor BBQ at least 185 times (boy will he be steamed). 3) Assist in killing 370 people. 4) Personally cap the intel or capture point 37 times. Or be on a team that caps the intel or capture point 74 times. 5) Personally defend the intel or capture point 123 times. Or be on a team that defends 370 times. 6) Build a complete engy setup (sentry, dispenser, teleporters) 132 times and hope they're blown up by the enemy.
Whew, I think some of us are going to be busy!!
|
|
|
Rank: Administration

Groups: Administration, Newsie, Registered User, Server Admin
Joined: 11/6/2007 Posts: 987 Points: 783
|
I think the idea behind the changes would be to ultimately normalize the point structure. Yes, there'd definitely be some hard times as point totals for the top players plummeted, but the point is that as time went on, you'd start getting points back as well and overall it'd reach a balance and be harder for the kind of imbalance we see in the graphs to reproduce itself. Giller [GwDR wrote:I don't really care about my rank....as it is now I will pretty much stay where I am at unless somebody gives up playing.....there are 20k-70k spreads between top ranks....Fuck trying to bridge that gap. And that's exactly what this plan is trying to remedy. The idea that you'd have to get enough points to be in the top 20 just to go from #17 to #1. HolyJaw wrote:In the end, everything Endyss said.
|
|
|
| Users browsing this topic |
|
Guest
|