PROPOSAL - Future Ongoing TG Competition (Arcade/MAME)

  1. 11-15-2020, 09:38 PM
    One additional thought just occurred to me which could effectively impact the spirit of the competition...and it is a situation that I dealt with on several occasions in the past. The problem is what if anything can be done about it.

    ******************

    "Spoiler" participation.

    While only eligible players would be ranked for the championship (i.e. one validated score per title), it is possible that a player skilled on one single title can put up an impossibly stellar top score, and minimal participation on the other nine. Having this happen across all ten titles would create "score clustering" for many lower-ranked scores.

    Imagine (in theory) a title has a grandmaster-class player submit a validated 1M score while the next closest players all cluster between 100-200K, thus no more than 10-20% per title. Have this happen across all ten titles then out of a max 1000% the best one could hope for is an aggregate 200% or so...unless one of the top clustered players possesses one of those stellar out-of-reach scores.

    "Score clustering" is to be expected in some but not all titles. The dynamics of certain titles make this inevitable. Time-based challenges limit point-pressing to a certain extent, but stage-based encourage extreme point-pressing without the time limit. Each suggested criteria carries with it a different risk-vs-reward outcome. Take "Galaxian"...in a 15 minute challenge do you go for boards or play slightly slower as non-stationary enemies are worth more points ? But in a stage-based threshold then you clearly maximize each stage and avoid shooting stationary enemies to the maximum extent possible while simultaneously going for max command ship points per stage. The chance of "clustering" is actually greater in a stage-based challenge in some cases depending on the dynamics of the title.

    "Missile Command" is the opposite. In a time-based challenge the missiles/planes/smart bombs can only come out so quickly per stage no matter how efficiently you pick them off. Thus at the end of, say, 15 minutes most experts who lasted thru the end will be within just a few stages of one another, although the points will very in terms of efficiency along the way.

    So the concerns of both "score clustering" on some titles exists as well as "spoiler scores", and specific to the latter, what to do about that as it is impossible to know in advance to some extent whether that is the intent of the submitter. The only solution I can come up with is to avoid title inclusions which carry with them the possibility of a runaway, nearly untouchable top score in the hands of an elite 1-3 players, in terms of disparity between all other players.

    The Atari 2600 "Time Deca" from nearly 20 years ago educated me in the do's and don'ts of such an event. Score (time) clustering was extremely widespread on most titles, and the few that were whole-second-based ended up being the determining factors of the championship itself. Half of the titles had clustered results in a very spirited competition, but those were over-shadowed by the inclusion of "Breakout", "Superman" and other whole-second based titles...especially the ones with the least amount of seconds to complete the performance.

    https://www.twingalaxies.com/showthr...-Final-Results

    It was not included, but "Firefighter" which had a then-record of 7 seconds and a second-best record of 8 seconds is a major indication of how this failed...a mere one whole second in disparity was a huge percentage difference which would have dwarfed that of the titles that were dual-decimal-based in the deca event. As it was, "Breakout", "Superman" and "Adventure" alone ended up determining the final ranking this event.

    The lesson to be learned from here is to craft a decathlon event where it won't come down to keeping pace on a few titles and focusing on a the few with the greatest point disparity potential.

    That stated, thanks for all the opinions and nothing at all is set in stone, especially the "Final Ten" titles and the associated settings/criteria.
  2. 11-15-2020, 10:44 PM
    I love the idea. I also agree with the point made that a true consensus on ten titles would be hard to reach. Here's my input:

    In general, aside from truly marathonable games like Robotron, what's the point of limited-life challenges? If this is a never-ending competition, without the need to squeeze in 10 performances and verifications for many players into a week or whatever, then what's the harm in letting players "play it out"?

    Donkey Kong:
    Time-based challenges don't really work here. Running the boards/speedrun games will always score higher than point-pressing games in that scenario. A 60 minute score for a speedrun would be in the low-mid 600k range while a point-pressing/WR game would be in the mid-300k range and neither approach would result in a killscreen. Restricting DK to time-based runs doesn't do anything to separate the great players from any other tier of player.

    Moon Patrol/Tutankham:
    There really isn't any point-pressing in these games, so time-based or level-based challenges also seem like a bad fit. Eventually you'll get many players who can last X minutes at Moon Patrol or get to Level X in Tut and their scores will all be clustered together. Single-life would be a terrible fit for Moon Patrol but could make more sense for Tut...but I don't like the idea of non-marathonable limited-life challenges anyway.

    I think a combined MAME/Arcade scoreboard would be the way to go, but if that ruffles feathers Jace/TG might consider 3 separate scoreboards: Combined, MAME, Arcade.

    If I had to choose from the initial list I would go with:
    Centipede
    Donkey Kong
    Galaga or Galaxian
    Jr. Pac-Man
    Missile Command
    Moon Patrol
    New Rally-X (swapped for Rally-X)
    Robotron
    Track & Field
    Tutankham
    Thanks RTM thanked this post
    Likes Barra liked this post
  3. 11-15-2020, 11:29 PM
    The scoring system needs to be talked about also. A percentage based system has good and bad and I've always liked the system as it made it so that if you knew one player could smash everyone then you didn't have to play that game as the points earned were so low that everyone got between 0-20 points. Using percent based method makes it so that some games are worth more and others are worth less.

    Now using the point based system of 100 for first and 99 for second makes every game important and makes it so that playing that game to extreme levels has less affect on the overall level of an elite player. Or you could also give 100 for first, 90 for second, 85 for 3rd, and then 80 with every score after that worth 1 less point, so 6th place would be 78. It's nice to make a bonus for a top 3 score as it will keep a close title very competitive and fresh.

    Time, number of lives, stages, max, are all things that need some discussion. I like the idea of time, short games level the playing field for most players and makes it so that the games aren't crazy long or make it so that you only make a few attempts at it. I guess you could always try out all the ideas and see what most players play or like best. Either way it will take time to gather some data, input, and discussion of all ideas and then a decision.

    As far as MAME, FPGA, Arcade, I'm ok with it. I've made my feelings known on these platforms. They will play similar enough and fair enough that once tested or discussed will become more clear. I'm yet to find a platform on the games that I play to say that there is significant difference that they need to be separate. There are instances that have been noted or found but once the list is put together of games we can either test it or discuss, talk to some of the pros about the game in question.

    Robert it's great to have you back creating fun and interesting challenges.
    Thanks RTM thanked this post
    Likes Blackflag82, evan04 liked this post
  4. 11-16-2020, 03:28 AM
    Xelnia said - "In general, aside from truly marathonable games like Robotron, what's the point of limited-life challenges? If this is a never-ending competition...then what's the harm in letting players "play it out"?

    Donkey Kong:
    Time-based challenges don't really work here


    Moon Patrol/Tutankham:
    There really isn't any point-pressing in these games, so time-based or level-based challenges also seem like a bad fit. Eventually you'll get many players who can last X minutes at Moon Patrol or get to Level X in Tut and their scores will all be clustered together. Single-life would be a terrible fit for Moon Patrol but could make more sense for Tut...but I don't like the idea of non-marathonable limited-life challenges anyway.


    I think a combined MAME/Arcade scoreboard would be the way to go, but if that ruffles feathers Jace/TG might consider 3 separate scoreboards: Combined, MAME, Arcade"

    RTM REPLY - as I had hoped a community-based proposal would spur ideas and flaws/nuances not initially considered or that were glossed-over. My initial concern was that marathon performances, even those in excess of an hour, could work against the adjudication process, but it seems apparent that if the right titles are selected in which performances could exceed an hour that would be fine...but there has to be a reasonable limit as I do not envision much community interest in adjudicating performances in excess of 24 or even 12 hours on a recurring basis. Should there be an unofficial "cap" on titles for consideration ? And by that I mean not necessarily a hard cap of 6 hours but performances where a 1st life, limited life or full game is highly likely not to exceed the cap hours ? That would make competition less clustered while still making adjudication less of a chore. And if a cap on hours seems logical, what should the cap more or less be ? That will of course still exclude some titles ("Tron") while allowing many more candidates come into the picture for final determination.

    "Donkey Kong" need not even make the final cut, BTW...it simply was a strong first contender due to its mainstay prominence, but not essential. Personally I'd rather have "Tempest" in the mix than DK.

    Single life "Tut"...I like that !! The title is such that even an expert still does not have a cakewalk thru stage 16 so a lot of opportunity for point disparity and avoidance of score clustering exists...good thinking.

    Combined-vs-Arcade-vs-MAME might not work as over time the Arcade possibility will no longer be feasible and the rankings will remain static, forever etched in time. Understanding that the classic era is already 36+ years old, and that the goal is a never-ending competition, that possibility will become problematic in just a few short years. I'm more of a proponent of a hybrid of the two rather than isolating arcade and MAME.

    Redelf said - "The scoring system needs to be talked about also. A percentage based system has good and bad and I've always liked the system as it made it so that if you knew one player could smash everyone then you didn't have to play that game as the points earned were so low that everyone got between 0-20 points. Using percent based method makes it so that some games are worth more and others are worth less.


    Now using the point based system of 100 for first and 99 for second makes every game important and makes it so that playing that game to extreme levels has less affect on the overall level of an elite player. Or you could also give 100 for first, 90 for second, 85 for 3rd, and then 80 with every score after that worth 1 less point, so 6th place would be 78. It's nice to make a bonus for a top 3 score as it will keep a close title very competitive and fresh.


    Time, number of lives, stages, max, are all things that need some discussion. I like the idea of time, short games level the playing field for most players and makes it so that the games aren't crazy long or make it so that you only make a few attempts at it. I guess you could always try out all the ideas and see what most players play or like best. Either way it will take time to gather some data, input, and discussion of all ideas and then a decision.


    As far as MAME, FPGA, Arcade, I'm ok with it. I've made my feelings known on these platforms. They will play similar enough and fair enough that once tested or discussed will become more clear. I'm yet to find a platform on the games that I play to say that there is significant difference that they need to be separate. There are instances that have been noted or found but once the list is put together of games we can either test it or discuss, talk to some of the pros about the game in question.

    RTM REPLY - the percentage-based scoring system was the initial thought, but nothing is closed for debate. I agree that with score clustering comes the inevitable percentage clustering...that's what happened with the Time Deca for several selections. And you hit the problem right on the head that I had with that Deca...some titles were more advantageous to play for purposes of outcome than others.

    With that in mind, even a 100, 99, 98, etc point-based scale causes inevitable clustering where a single point increase on a title can change the overall ranking. That said, your second suggestion is more of a "sliding scale" based method. The problem with the 100, 99, 98 method is what happens when 101 players become eligible...and then 102, etc ? The problem with "sliding scale" is that the scale also needs to be adjusted based on volume of participation. If only top scores or the top 10%, 25%, 50% of scores, etc have a value attributed yet the ones below are worth practically zero or even zero, that dissuades competition once one of the titles reaches that point where the elite performances far outrank the others, something I was initially hoping to avoid.

    That's the trade-off between allowing "score clustering" and finding ways to prevent it...one headache vs another. An "adjustable sliding scale", however, that might work. Establish the values for the first 50-100 scores and have the scale adjusted to compensate for further participation. What that means is that increased participation still assigns values to every performance based on the top performance but it also diminishes the distance between performances. Let me explain.

    Suppose for the first 100 scores you have a scale of 100 for 1st, 98 for 2nd, 95 for 3rd, 90 for 4th-5th, 80 for 6th-10th, 75 for 11th-20th, 70 for 21st-30th, 60 for 41st-50th, 50 for 51st-60th, 40 for 61st-70th, 30 for 71st-80th, 20 for 81st-90th, 10 for 91st-95th and 5 for 96th-100th...or something like that.

    Now suppose we have one day 1000 scores (for illustrative purposes). The scale of rankings must be adjusted to cover 1000 possibilities, not just 100, so is it still fair to have 75 for 110th-200th place or do we start tweaking such as values from 79-71 spanning the before and after thresholds...where 110th-119th is 79 points, 120th-129th is 78 points down to 191st-200th is 71 points ?

    I'm open to suggestions for both initial and adjustable sliding scale methods/thresholds as well as alternative viable ranking methods as this is a work in progress.

    Next you mentioned shorter performances. There's a place in such a mixed competition for performances of all durations save for marathons and ultra marathons as they quickly eliminate score clustering at the top though create MASSIVE score clustering at the bottom (using "Tron" again envision 2 performances at 12-15M of more and the other 99% all being no more than 8% of the total at best at 1M or less, with most being less.

    Some titles by their very nature will be clustered ("Track and Field". "Pole Position") if included. Percentage-based tracking creates ranking clustering but adjustable sliding scale eliminates that...the trick is selecting the right scaling method.

    I agree that the way to go is allowing arcade and MAME to co-mingle here as this is a unique event proposal.

    Robert
  5. 11-18-2020, 12:23 AM
    yay classic only while late 80's and onward arcade score attack is literally forgotten

    :)
    stg never die. 47 stg 1cc's
    Likes The Evener, Luigi Ruffolo liked this post
  6. 11-18-2020, 07:18 PM
    I'll just go through a few games a day and my thoughts on your list or proposed so far.

    Asteroids.
    Pros.. Very common, most people know how to play.
    Cons. Too many variants of the game, has 4 romsets all slightly different. Speed up kits and mods, so timed game is out.

    I would suggest Asteroids Deluxe in it's place, slightly less common, no known speed up or mods, 2 versions of the gameplay which can I believe can be set to the same on either version of the romset. Would be 5 lives setting as it is marathonable. Could also do 15 minute time limit but I'm guessing most players can't last that long. 15 minutes is around 125k or so.

    Berzerk.
    I would suggest going with 15 minute time limit, not all people can play this game even that long. Need to choose slow or fast bullets.


    Black Widow.
    Not very common but an ok game. Not a popular game among casual players. Game has a kill screen although nobody has every reached it without skipping or continuing gameplay. Most people top out in the 200k or so range so a nice range of scores for those that try a little harder.


    Carnival.
    Very uncommon. Top scores will be from 40-60k and games are of quick to med length. Good tourney game.

    Centipede.
    Great game, can be a really quick game with 3 minute, lots of replay value. 1 hour won't work as most players can't last that long. 15 minute is probably the way to go.


    Defender.
    Great game for tourney. Needs to be timed. I would use factory settings and 5 minutes. Scores would max out around 100k with 5 minute limit. If 15 minute used scores would range from beginner scores of 10-20k up to almost 300k.


    I'll end here. Yeah so far I would include the above games in an ongoing event like you have suggested. As more games are discussed though I would be tossing out 1 or 2 of these games.
    Thanks RTM thanked this post
  7. 11-18-2020, 09:51 PM
    Quote Originally Posted by Pearl2hu View Post
    yay classic only while late 80's and onward arcade score attack is literally forgotten

    :)
    Excellent point! Do you have any suggestions of some great 90s titles? I spent all of my time in the 90s playing Street Fighter II and its future variants, Trog, and a lot of Dragon's Lair II when it came out. :)
  8. 11-19-2020, 03:12 PM
    it depends on what exactly the competition is aiming for, is it a timed thing? life based? who knows
    stg never die. 47 stg 1cc's
  9. 11-19-2020, 03:14 PM
    If this can get up and running it would be neat to then add additional year long contests on the bigger gaming platforms.
  10. 11-20-2020, 09:30 AM
    Quote Originally Posted by redelf View Post
    I'll just go through a few games a day and my thoughts on your list or proposed so far.

    .....

    I'll end here. Yeah so far I would include the above games in an ongoing event like you have suggested. As more games are discussed though I would be tossing out 1 or 2 of these games.


    RTM REPLY - Thanks !! I'll start collating all the responses and updating the initial potential roster of 37 later today. Great ideas on these specific titles :)
Page 2 of 5 FirstFirst 1 2 3 4 ... LastLast
Results 11 to 20 of 50
Page 2 of 5 FirstFirst 1 2 3 4 ... LastLast
Join us