This topic contains 54 replies, has 0 voices, and was last updated by  Deathbal 9 years, 4 months ago.

  • Author
    Posts
  • #52079

    BigBear
    Participant

    As much as I enjoed reading the latest INTERWEBZ DRAMAH direction this thread has taken, I feel I should make an effort to try and give some background info and possibly some answers to the questions asked by Rommel. This as I started this thread while still being part of the admin crew.

    @rommel wrote:

    Who is heading up this project?

    I was the one that took the initiative and since I quit without discussing the continuation of this topic with my former admin colleagues it’s not strange that it was sort of left blowing in the wind. It would seem to me that Outer is trying to pick up the pieces.
    @rommel wrote:

    What is the goal, specifically?

    The initial goal was to (hopefully) find a better balance between kills and round wins.
    @rommel wrote:

    What conditions prompted the changes?

    It has been informally discussed and mentioned in several topics over time. POM’s request was perhaps the spark, but it was long in the making.
    @rommel wrote:

    How were the changes expected to achieve the goal?

    Considering the above mentioned goal I would say that tampering with the kill/win ratio is self explanatory.
    @rommel wrote:

    What was the exact result of the changes?

    Getting quality feedback has always been a problem within this community. That is in part due to imperfect routines of gathering it but also because only a small fraction of the players involve themselves in forum business on a regular basis. If you are looking for statistics or a nice diagram you can forget about it. 😉
    @rommel wrote:

    Why did the original changes not produce the results desired?

    If the answer to the problem was known before trying it out there would be no need for a debate to start with. Then a simple declaration that “These are the new score settings PERIOD!!!” would suffice.

    I haven’t played at all lately so I can’t provide any useful input on the changes, but based on past experiences a 2.5 kills per round win ratio makes a lot more sense to me than the old score settings. As for assists that was not something I even considered when I started this thing and presently I have no opinion on that subject.

    #52080

    Outer
    Participant

    Hey BB, nice to see ya again talking. Maybe im not that good at posting long posts Romms. I answered with what i knew, maybe things got not clear because was very early in the morning here 3 AM and i was at the edge of sleep…………………….. so i’m pretty sure i messed up my english #-o . Not posting more at that hours for me! 😉

    As BB says, yep im trying to build my knowledge on this issue to try to do a good admin job for you guys, after BB left the house 👿 . Anyway I was joining the servers to see players reactions about the new scoring. Some didn’t notice at all that this has changed, others took it well and seems to be comfortable with the settings.

    Yes is a shame that there are just a few people that post on a regular basis and that are involved actively in the community projects and don’t only go to check out the stats and that’s it. Maybe a mistake from our part too… need to encourage people to post like Rommel 😛 (insane).

    Rommel, lets not fall into emotion(both of us) and try to point out the errors of the other… its fun and all but that wont help us on the “debate”.

    You know its difficult (everyone has its own lives going on) to have data here data there to show you that the changes being made are correct. Obviously they are made by assumptions and trial and error. But not all is left to luck and randomness…take out some ratios, see if they fit better, set them on the servers, that’s the methodology of this.

    About the assists, and i know Rommel, assists can define a game and are important. What we wanted was to allow the score not to be only multiples of 10, we wanted to approach to a more tight scoring system that could left apart players up to 1 point of lead. Why the assists you may ask again… well unfortunately and is a request i posted on the feature request & bugs forum is that we dont have many options to change the score. I pointed out that it would be cool to earn some points for hitting a shield, so assists would not be changed at all. The options are kills, wins, assists, lives and cash u have. We could have tweaked lives and cash but we didn’t see the point, these options are less far important in main server. Imagine if we give point for every 100 money units a player has… ](*,) . In the absence of score options we took assists.

    On the other hand for a mid to large term project we could check the logs, what you proposed, and take out more strong conclusions about how the game is affected by the different factors, wins, kills, assists

    For now its gonna be tweaking and its true…

    Outer

    #52081

    Rommel
    Participant

    Hi Outer:

    I’ll keep this one short and tie up the loose ends.

    The main issuse with making these changes for me, and I suspect for others as well, has been the inability to mentally calculate if the leader can be caught. Just as I was getting proficient with the new scoring system, it changed again and replaced a multiple of 10 with a 9. It really doesn’t matter to me about the ratios because I play based on the scores, not on the ratios.

    A test server using 12 to 24 bots running 24 -7 as a test platform might help speed up the process and allow you to ensure that you have tried every plausable combination without introducing a flurry of ineffective changes into the game. This would be a much less obtrusive way to do trial and error testing for a Customer [sic] than to do it on the companies main file server during business hours.

    It seems strange to me that after seven years of operation, a suitable ratio still hasn’t been determined but I feel quite certain that you must be getting close by now. Of course this assumes that you aren’t trying settings that have been tried and deemed unacceptable in the past. I wish you all much success and God speed.

    Sincerely,

    Rommel

    #52082

    Outer
    Participant

    @rommel wrote:

    Version 41, records the final scores in the log file. Thank you Gavin, that was a terrific idea. This addition provided the admins an easy way to compile the data required to perfom analysis on the scoring.

    Does this log display anything to you? I checked out the server settings for logging and they are turned off. Bad thing for our purposes.

    #52083

    Rommel
    Participant

    Hi Outer:

    @outer wrote:

    @rommel wrote:

    Version 41, records the final scores in the log file. Thank you Gavin, that was a terrific idea. This addition provided the admins an easy way to compile the data required to perfom analysis on the scoring.

    Does this log display anything to you? I checked out the server settings for logging and they are turned off. Bad thing for our purposes.

    I’m almost certain that the server is creating log files somewhere. Perhaps one of the other admins can help you find them.

    For rapidly finding and stripping out the scores, I search for the phrase: Final scores.

    This is from my serverlogxxxxxxxx.log file

    Wed Jul 16 16:07:47 2008 – [banner] : “R12 is the overall winner!”
    Wed Jul 16 16:07:47 2008 – Final scores


    Wed Jul 16 16:07:47 2008 – Bullseye – 1195 (20917$ 4K 6A 2W 1L)
    Wed Jul 16 16:07:47 2008 – Tura Satana – 714 (8903$ 2K 9A 1W 0L)
    Wed Jul 16 16:07:47 2008 – michaelmr. – 1085 (5768$ 1K 4A 2W 0L)
    Wed Jul 16 16:07:47 2008 – Pampaduor – 707 (12037$ 4K 7A 1W 0L)
    Wed Jul 16 16:07:47 2008 – Nighty – 1192 (23178$ 3K 6A 2W 1L)
    Wed Jul 16 16:07:47 2008 – chinasky – 1091 (5987$ -1K 7A 2W 0L)
    Wed Jul 16 16:07:47 2008 – BACEK – 1060 (5947$ 0K 4A 2W 0L)
    Wed Jul 16 16:07:47 2008 – Rommel – 1160 (8818$ 2K 5A 2W 0L)
    Wed Jul 16 16:07:47 2008 – O.G. – 1140 (5890$ 3K 4A 2W 0L)
    Wed Jul 16 16:07:47 2008 – Merlin – 564 (5000$ 0K 4A 1W 0L)
    Wed Jul 16 16:07:47 2008 – Val Chaon – 641 (6220$ 3K 4A 1W 0L)
    Wed Jul 16 16:07:47 2008 – diyadokeon – 661 (5031$ 4K 3A 1W 0L)
    Wed Jul 16 16:07:47 2008 – R12 – 1584 (22818$ 4K 2A 3W 1L)
    Wed Jul 16 16:07:47 2008 – NoName – 1180 (27193$ 3K 7A 2W 1L)
    Wed Jul 16 16:07:47 2008 – Gomez – 662 (5822$ 2K 6A 1W 0L)
    Wed Jul 16 16:07:47 2008 – SLIM – 1212 (21765$ 5K 3A 2W 1L)
    Wed Jul 16 16:07:47 2008 – ninja laptop – 678 (8302$ 3K 6A 1W 0L)
    Wed Jul 16 16:07:47 2008 – xox – 1231 (20977$ 5K 4A 2W 1L)
    Wed Jul 16 16:07:47 2008 – Wooky77s – 1122 (22610$ 2K 4A 2W 1L)
    Wed Jul 16 16:07:47 2008 – Nobody – 1115 (17688$ 4K 1A 2W 1L)
    Wed Jul 16 16:07:47 2008 – Rommel(2) – 645 (31108$ 2K 2A 1W 0L)

    Yes, the bots beat me, they often do.

    You might notice that it wasn’t the kills that determined the winner, it was the wins. The scores were still pretty close in spite of the fact that I use a much different scoring system than you are currently using. What do I use? Awww gee, thanks for asking.

    Round win = 250 points
    Kill = 25
    Assist = 15
    Life = 250

    So a round win in this example is worth 500 points and a kill is worth 25.
    In order not confuse the issue, I reset the lives to 1 instead of my normal 2.

    As you can easily see, my assist value is 60 percent of the value of a kill and the assists in this game were a big factor in the scoring.

    Best wishes,

    Rommel

    #52084

    Rommel
    Participant

    Hi Outer:

    In light of how much time we’ve saved ourselves today … and figuring you hadn’t reserved all that extra reading time I suggested, I thought I would reformat the output to make it a little easier for you to read. Since the BBS will not display it properly without excessive editing, I have attached it as a text file. This should be much easier to decypher than what I posted earlier.

    Something that I failed to mention earlier is the cash awarded. In reality, there are many things that contribute to the final score and cash awards are something that hasn’t been mentioned.

    My cash awards are as follow:

    Hit Points = $ 250
    Kill Points = $ 750
    Multi-Kill = $ 500
    Win Points = $ 5,000
    Life = $ 5,000
    Assist = $ 250

    Something else that effects the score is the type of weapon used. A player making 10 single kills with baby missiles earns substancially more cash than a player making 10 single kills with nukes. As all should know, a player with abundant cash is better suited to survive the next round.

    In closing, I’d like to quote General George S. Patton – “No bastard ever won a war by dying for his country. He won it by making the other poor dumb bastard die for his country.”

    Contrary to what has been posted here by others, the object of war is not killing, it is surviving. You may have to kill to survive but you don’t have to die to kill. At least I don’t, not always.

    Best wishes,

    Rommel

    #52085

    Rommel
    Participant

    Oooops.

    Assists: Compare the results for Chinasky to those of R12. It is impossible to say without recording the action but it seems reasonable to suspect that one of Chinasky’s seven assists could have helped R12 to win the game. However, since the assists are 60 percent of a kill, Chinasky needed only to survive the final round to win the game.

    Perhaps my cash awarded for assists is too low and that is why ChinaSky had no money going into the final round.

    I hope that this helps to illustrate that there are a great many things to be considered and that the assist value is one of them.

    #52086

    hobbesme
    Participant

    @outer wrote:

    assists can define a game and are important. What we wanted was to allow the score not to be only multiples of 10, we wanted to approach to a more tight scoring system that could left apart players up to 1 point of lead.

    The concept of adjusting the assist score so that scores might be differ by only a single point would make for very competitive games.

    Unfortunately, adjusting the point value for assists to 9 points will not substantially contribute to games being separated by a single point very often. It can happen, but it will NOT occur frequently, as several of you hoped it would.

    More often than not, close player score differences will be +/- 4 points or +/- 9/10 points — very rarely +/- 1 point.

    #52087

    Deathbal
    Participant

    True hobbes, but the possiblity is there. On several occasions I have been losing by 2 points in games. It is working out fine.

    #52088

    hobbesme
    Participant

    @deathbal wrote:

    but the possiblity is there. On several occasions I have been losing by 2 points in games.

    A 2-point differential is the lowest most likely difference :

    100 points (1 win) vs. 98 points (2 kills + 2 assists)

    A one-point differential requires quite a few assists :

    100 points (1 win) vs. 99 points (11 assists)
    81 points (9 assists) vs. 80 points (2 kills)

    All other 1-point differences are based on these two differences adjusted by additional wins/kills/assists.

    Personally, I think that a 4:1 kill-assist ratio is OK but a 40/9 ratio may deprecate the value of assists. I don’t know that I agree that assigning 9 points to assists as a supposed means to make games closer in score outweighs further reducing the effectiveness of an assist.

    #52089

    Deathbal
    Participant

    I do see your point. Does the close scoring outweigh minimizing assist value? I truly don’t know. We can solve both problems by making them worth 11 😀

    I think the idea was to not have the closest score be by 10. It just seems to hurt more when you lose by 4, or 3…….kind of makes it more interesting.

    Remember, assists used to always be a 5:1 ratio and no assists were awarded for a resign. Now it’s almost 4:1 and you get an assist when people resign. So their overall effectiveness has been increased.

Viewing 11 posts - 46 through 56 (of 56 total)

You must be logged in to reply to this topic.