This is the bat signal - let's talk about TWDT

[Jones]

I think it's impossible for a bot to correctly measure skill. Eg. Someone pressuring a lot, forcing people into bad positions etc. while getting a subpar score.

[ogron]

You think Qan is going to design a complex Real Plus-Minus statistical system and code it into the bots for 5 different ships in a dying game with no TWD or TWL activity on which to even use it on? Do you have any idea how much work RPM requires? Every NBA season has 60,000 game segments and you have to run R-squared regressions through a special program. Features that were 10x simpler and 10x less time-consuming that I asked for to be coded into the TSL bot never materialized.

Your previous suggestion is looking for sample sizes larger than the entire TW population.

Nothing you posit is even remotely realistic. Therefore, it is not helpful. That leaves 2 conclusions:

1) You're not intelligent enough to realize these suggestions have no value because they're completely unrealistic
2) You want to hear yourself talk

Such a system will never be implemented. If it was, there aren't enough games to have good sample sizes. If there were enough games, many players don't practice anymore and it wouldn't be useful in judging them. And people would intentionally throw in TWD and TSL games to tank their rating for TWDT to get to a lower star level.

That leaves 2 realistic paths:

1) You average out a small sample size of non-experts (something that has repeatedly produced the worst ratings)
2) Someone judges others' skills and consults with ship-specific experts

The latter is the best realistic option. To that point:

you think 1 guy who knows everything (you) is qualified to judge skill in a game (...)

You're right, I do think that. And so do a lot of other people. Anyone who has won multiple championships as a TWL or TWDT captain is probably an excellent candidate as well. I once benched Diakka in a TWDT-B Final, one of the biggest names in the history of basing, for someone with no Finals experience named MR. HOMO, and won. Hulk, Zidane, Mythril, Ease and others have done the same. If you want to win, you have to be able to evaluate talent on an extremely granular level. I've been benched by people who liked me way more than the guy they benched me for, and the talent difference was small. Competitive people care about winning. None of them give a shit about popular perception.

but perceptions of skill are largely just public opinion.

That is absolutely fucking untrue.

You don't win championships selecting players based on public opinion. You win championships identifying more skilled players at a higher rate than your opponents. The best players and captains in this game have played THOUSANDS of games with and against most of the players in TWDT.

That you think I would do ratings based on public opinion shows how little you understand about this game, or the veterans who excel at it. You want to know what the average thought process is for each player? Here you go:

Why is x player in shark an 8 and not a 9? Well, I'll tell you so. Because one of the top 3 sharks to ever play the game told me so, and he's played with and against the guy around 500 times. He reps too high in the cram, he's letting the enemy team crawl up the right side and establish position to clear out the weakside spider on the left of the cram, letting the enemy terr climb up and the next shark to attach and rep, pushing our team in the cram out, and letting them break.

Because he doesn't know all the burst patterns, and isn't repping bursts on the flag for the underspiders, and he isn't shifting right off attach to get to the right spots in time do so. He chokes in the clutch and makes bad mines in desperation. He's getting his 3 reps off, does okay in the cram, doesn't make too many mistakes, is a decent 2nd shark paired with a better player, but he focuses too much on trying to shrap the enemy terr and not enough clearing paths for his spiders to push. He isn't great against laggy euros either and struggles more with the timing on them. He doesn't watch his death screen closely enough before attaching and sometimes reps when the terr is close to enemy mines/walls.

But he's very active, has good reflexes, can be coached, and makes enough right plays that he's too good to be a value 7-Star. So he's an 8.

And that's the ship and position I understand the least.

I've played so much spider I can tell you which spiders intentionally let you shoot the enemy sharks first so they can get the last kill and boost their stats, which will camp and anti-camp in the cram and how fast their setup is, which spiders change their ship-speed when firing at sharks to change their bullet speed to force double reps, who knows how to push behind friendly sharks, who camps too much, who dives too much, who knows the burst patterns and who doesn't, especially underspiders, which as a terr I will change my bursts depending on the spiders playing under.

I don't burst under if Stayon/Acidbomber is there because he knows every pattern by heart, always spots the bursts ahead of time, and plays around them. Mikkiz is going to set a port high on go and try to port trap you if he loses the early battle. He won't burst the side holes at the entrance when breaking cram, so spiders can sit there freely without needing to worry about conserving energy to re-attach.

Zidane can tell you just as much about Javs.

Mythril can tell you which warbirds stand their ground, which WBs don't cross, who knows how to lure, who spots vulching opportunities, who shifts to dodge, who spins in place to dodge, who will actively fight for the corners of the enemy lines so you don't get flanked, who will intentionally clip through a rock to abuse the client, who will bounce off a rock when chasing to surprise you and attack, who has the best energy management, who can perform in the clutch, who chokes, who is better in 5v5 situations and is the best rockshot or radar-player, who performs better in the later stages when its 2v2, who are the best 1v1 duelers, who knows how to stay alive on 9, who knows when to hunt and when to just play the game out and not force the issue, etc.

The notion that you think it's just "public opinion" and some bullshit popularity contest, and not a bunch of experts with 10,000+ hours in multiple ships who have several hundred data points on each TWDT player when evaluating them, shows that you have no idea what the fuck you're talking about.

[Falconeer]

Falconeer

First off, I strongly urge you to go see a psychologist. Your constant, unending forum spam seeking attention and desperate need to sound intelligent and gain some sort of validation is not normal.

I ran 5 TWDT seasons. I have more experience than anyone running leagues. That's not an opinion, that's a fact. When I say "averaging out doesn't work well", it's because it has been tried. I'm not busting out MLA citations and links just because you have a tumor pressing against your brain.

Oh damn it looks like someone's ego was hurt and they are reciting their resume to me. I didn't ask for your qualifications in a 2D game.

Your Wisdom of the Crowd video has sample sizes of 1,000 to 10,000 people. It took me five weeks of hard work just to get 11 people to do the one TWDT season where we used averages (throwing out troll answers (i.e., Dreamwin rating every other cap a 10/10/10 to hurt their ratings). I believe I used the interquartile mean of the non-troll answers, as that seemed to make the most sense with a small sample size. I didn't do this because I thought it would produce the best ratings. I did it because I was tired of the constant whining about ratings. After the averages season was widely panned, everyone was much happier with me going back to just doing them myself.

I agree averages work best with 10k people, otherwise it's better to use the geometric mean. But you didn't say that, you just said averages don't work at all... you think 1 guy who knows everything (you) is qualified to judge skill in a game, but perceptions of skill are largely just public opinion. Which is why i suggested someone like qan actually make a bot (which would have to be designed properly) and from there veterans can use the information the bot gives to evaluate proper ratings while also using some sort of rubric/evaluation system (if you designed that system then great, pass it around don't keep it to yourself). Using someones 'gut' feelings or unannounced thoughts to make ratings is what causes that issue of 'someone elses criteria for 9/10 player will be different from yours'.

Don't worry i don't need to sound smart, i care less what you or anyone thinks of me, i'm a faggot remember =)

[ogron]

Falconeer

First off, I strongly urge you to go see a psychologist. Your constant, unending forum spam seeking attention and desperate need to sound intelligent and gain some sort of validation is not normal.

I ran 5 TWDT seasons. I have more experience than anyone running leagues. That's not an opinion, that's a fact. When I say "averaging out doesn't work well", it's because it has been tried. I'm not busting out MLA citations and links just because you have a tumor pressing against your brain.

Your Wisdom of the Crowd video has sample sizes of 1,000 to 10,000 people. It took me five weeks of hard work just to get 11 people to do the one TWDT season where we used averages (throwing out troll answers (i.e., Dreamwin rating every other cap a 10/10/10 to hurt their ratings). I believe I used the interquartile mean of the non-troll answers, as that seemed to make the most sense with a small sample size. I didn't do this because I thought it would produce the best ratings. I did it because I was tired of the constant whining about ratings. After the averages season was widely panned, everyone was much happier with me going back to just doing them myself.

That was the goal, and it was a resounding success.

Here's the thing: Your video and opinion have NO RELEVANCE. The types of sample sizes being discussed are not possible. You're just posting things to try to seem smart and knowledgeable. It has no useful application to TWDT.

Moreover, in the video you linked, Galton's experiment gauged the wisdom of the crowd in guessing the weight of a butchered ox AT A COUNTY FAIR. As the presenter states, it was an event "where people knew quite a lot about meat and what that actual quantity of meat would come to". It is weighing a group of experts on a very simple task. Most Subspace players are NOT experts in all three leagues, let alone their own ship. Secondly, of the few experts there are, only a small percentage of them have spent significant time evaluating a vast swath of other players. Even fewer are willing to help TWDT. I can count on one hand the amount of people who can tell you the difference in skill between Ease and Apok in warbird or Cape and Zizu in terr. Yet the difference is significant. It takes thousands of hours of competitive play to know the difference between an empty stats Spider putting up 120-60 and a true difference maker who helps you win games. And that's 1 of 5 ships.

You can keep blathering on about the geometric mean and other Wikipedia entries like Tower, it doesn't help the league in any way.

The closest thing to Wisdom of the Crowd we've had is the 50-60 opinions you hear after a season on how the ratings and season felt. The consensus was the best ratings players have ever had was when I did them myself with the help of 4-5 experts. The worst were when they were averaged or someone else did them. It doesn't matter if you average out 10 people or 1000, because IT DOESN'T MATTER IF THEY'RE NOT EXPERTS.

Get that through your fucking head.

1) We don't have the sample sizes needed so your suggestion is of no value
2) If we did have a large enough sample, it wouldn't matter because we don't have enough experts, so your suggestion is of no value

And BTW, all the "theoretical" data collection you're suggesting someone write code for into the bots, I've already done. I have something called APM (Adjusted Plus-Minus), equivalent to RPM in the NBA, on multiple ships across thousands of games that helps form an objective opinion on each player as an isolated variable, which I have used in rating players for TWDT. Even then, it's somewhat noisy, given the sample sizes. RPM works best across 2 NBA seasons (or 164 games), and most TWL and TWDT seasons are 7-14 games. Most veteran players have between 50-200 games logged in the pros, so mileage varies.

The only thing you're contributing to is people's migraines as you endlessly spam forums desperately trying to impress upon people that you're intelligent.

[Falconeer]

There are always going to be haters, but the feedback I've received has been along similar lines. Like I said, I think the best ratings solution is consulting the best players in every league, and then, at the end, smoothing out the final ratings by yourself so they're consistent (Stayon and Zidane might have different ideas on what constitutes a 9 in jav, and averaging their ratings out is stupid.)

The absolute worst ratings we ever had while I ran TWDT, that got the most complaints, was the one season where I let all 8 captains do the ratings and averaged them out. They were horrid. Averaging out doesn't work well. All it does is absolve the Head Op from blame. TWDT Head Op needs to be able to take the hate and be accused of bias, and just do the best ratings possible. (Unless they're incompetent, in which case someone else should do the ratings or perhaps averages would be less damaging.)

Clause.... your opinions are just filled with fallacies and ignorance... you never back up anything you say. "Averaging out doesn't work well." Do you know that the laws of mathematics and the universe basically disagree with that statement you made? watch this: https://www.youtube.com/watch?v=n98BhnwWmsc and when you are done watch this: https://www.youtube.com/watch?v=iOucwX7Z1HU

Averaging or the wisdom of the crowd is the most accurate representation of data. The issue is you need 10,000 people for the most accuracy. With something that has a lot of bias like estimating someones skill in a game which is basically influenced heavily by popular opinion that formed over time of that persons skill, you would probably use the geometric mean.

I still think you need the person being rated to rate themselves. Completely ignoring the person who is actually playing, has the most facts about how they play since they are present with themselves all the time and you maybe saw them play 1 game from spec once... ignoring this is retarded. Get a bunch of people to rate the persons skill, including the person themselves, and then get one of the averages.

Even better is to base it on facts mixed with opinion and not just full on opinion. Make a bot that collects all measurable data and accounts for how a pro might play on the shittiest team imaginable. Anyone who plays just as good as that pro on a losing team would be top rated, despite their win/loss being 0-20. Individual skill does not = wins in a team game where your team is failing. Then take those bot ratings and have the player, the most skilled people in the game, and a third party evaluate how accurate the bot is, and get the average.

Someone can also make a sheet that explains what a 10, 9, 8, 7, 6, and 5 would be. What are the criteria you judge? rate them all separately add them up and see if it gives a 9 or 10. This way everyone is using the same criteria. For a 10 we expect someone to be near perfect in every criteria all the time. For a 9 they slip up once in a blue moon? for an 8 they are usually very consistent, for a 7 they sometimes mess up in our list of criteria, and so on. Make these statements, find the criteria being judged, and send the sheet out to the people doing the ratings. Someone can have a 7 dodge, a 5 aim, but have a 10 out of 10 flagging, or team ability, and suddenly they are rated an 8.5..

[mtine]

Merry Xmas u filthy animals!

[zidane]

keep crying doofus lol

[ogron]

Like you said yourself, the best it works when more than one person rates. Preferly spread over the old squads imo. I dont know who plays anymore ,but i am sure someone like commodo could help out instead of thunder only...

Beggars can't be choosers. Fiercers, Dicers, and Pene people have never contributed to TWDT in great numbers. Thunder vets are the reason TWDT even exists. The 2013 was going to completely fall apart before Undercut agreed to start it (with K A N E eventually taking over).

The people who stepped up to cap TWDT 2013? Ease, Zidane, Ogron, Hulk, Kate Upton, Infrared, Skatarius, and Dreamwin. Almost all Thunder players.

I have repeatedly gone out of my way to include players from all the old squads, and mostly been rebuffed. I've ask them to captain, do ratings, etc.

Dreamwin, Commodo, Stayon, Geio, and yourself have been some of the few who have been willing to help. Have fun trying to get a response from Fierce starters. I've asked like 30 different Penetrate core players over the years, and beyond Trasher writing a column for 1 season and Porkjet doing some recordings, none have wanted to help. Dicers? Lol.

I think maybe showing some gratitude for the amount of time and effort Thunder players have put into TWDT instead of disdain might be appropriate. When other groups wouldn't step up to cap and do other stuff, those guys often reluctantly stepped in to help me because I begged them.

Not sucking your dick ogron but you seem to be extremely good at doing ratings, from a warbird perpective you were pretty spot on.
mythril thanks for addressing the multiship multicap issue. Picking last, last season ruined me

There are always going to be haters, but the feedback I've received has been along similar lines. Like I said, I think the best ratings solution is consulting the best players in every league, and then, at the end, smoothing out the final ratings by yourself so they're consistent (Stayon and Zidane might have different ideas on what constitutes a 9 in jav, and averaging their ratings out is stupid.)

The absolute worst ratings we ever had while I ran TWDT, that got the most complaints, was the one season where I let all 8 captains do the ratings and averaged them out. They were horrid. Averaging out doesn't work well. All it does is absolve the Head Op from blame. TWDT Head Op needs to be able to take the hate and be accused of bias, and just do the best ratings possible. (Unless they're incompetent, in which case someone else should do the ratings or perhaps averages would be less damaging.)

[Majorcrisis2]

if he was that elite you coulda picked him, i did pick last :thinking:

Yea tard i was gonna pick a 7 star wb while i could get Cape... When i needed Cape much more for base. You are a doofus man.

[zidane]

if he was that elite you coulda picked him, i did pick last :thinking:

[Attacks]

Not sucking your dick ogron but you seem to be extremely good at doing ratings, from a warbird perpective you were pretty spot on.
mythril thanks for addressing the multiship multicap issue. Picking last, last season ruined me
I will do whatever is needed cap, player w/e

[Majorcrisis2]

Holy moly ogron... First you say that last season's ratings were terrible and then zidane should do them. You realise that zidane was the major rater last season? Me and mythril had some input yea... That 7* wb vet zidane got in FA was all cuz zidane rated him and Henry ignored mythril and myself. We all know zidane always uses some kind of 'cheat' to win this stuff.

Like you said yourself, the best it works when more than one person rates. Preferly spread over the old squads imo. I dont know who plays anymore ,but i am sure someone like commodo could help out instead of thunder only...

[mtine]

I wanna see vys make the ratings again, just for shits and giggles. Guy is probably a stand-up comedian irl.

[Rasaq]

Can't say that I ever pushed to have a year long season of TWDT, but a few years back I did ask for the idea of a TWD style season of TWDT where captains could manage their rosters and trade players throughout the year. Rosters would be remained locked so no player could leave their squad. Could have players drafted to a squad on a yearly contract and once they served their contract they can either reapply to the squad (at the end of the season) or enter some sort of market. Like TWDT, the rosters would have an overall star system, so in order to p/u a free agent you would have to release a player with greater or equal value to acquire him/her. I'd love to see a captain have more responsibilities with their roster and be pushed to do more and be more active. This was something I suggested way before the game's activity started to dwindle so instituting this idea now would probably have little ambition.

[Falconeer]

I don't find player ratings accurate at all. It's usually bias and informed by personal feelings for the player, very rarely does someone objectively 'rate' someone, and often times it's just based on a reputation or public opinion that has formed about that players skill which is why even if you get multiple people to rate the same player and use an average it will likely be based on that public opinion and not actual intimate knowledge of their skill. In my opinion you should have the player rate themselves. Have a very skilled informed person (like zidane rate the player) and then have a third party rate the player. Then take the average of those three ratings. This only works if the person rating themselves is not being a douche on purpose... ie: cyclone saying he's a 10/10..

In DSB if trasher was a 9/10 (which is the rating i would give him), i would give myself an 8/10 and i would go further and say trasher is easy to kill, i've faced tougher opponents (10/10s). But the public opinion on my skill is like 6/10 at best, because i never play on stacked teams or teams i perceive will win like trasher does. If a team is bad, trasher doesn't even get in the game, he will sit in spec. That's where we differ, i get in to support activity and get people playing. Also skill level is dependent on various factors not just kill ability, or ability to stay alive, but choices you make while playing.

As for staff, it's not reasonable to expect one person to contact 400 people individually. I've been doing the same thing on steam, and it's a long drawn out process that takes days if not weeks, and lots of time. The reason i can do this is because i have no life. But most of us are in our 30s now, have jobs, have other commitments. People in staff are volunteering some time.... but not their whole life to the game. With that said we *NEED* someone in charge who is *STABLE* who is going to give you a fair shake even if they dislike you, who is not ban happy, and right now i see that person as Henry Saari. However, this doesn't mean he can't delegate and hire other league staff to contact players to join the league. See if you can work something out with him. I am perfectly sympathetic to peoples busy lives, what I am less supportive of is the fact that staff won't let people help.

I really want to see qan design a rating system that can be used as a baseline for player ratings; something that takes everything measurable into account as well as weighs skill based on the outcome of the game. So if you played amazing on a losing team you'll be rated high. These ratings then can be adjusted based on player input. So you can have a team of people maybe 5 people who have special login access on the site and can +1 or -1 a players rating out of 10. As well the player themselves can +1 or -1 their rating. And once you plus or minus someone you can't take it back that league. This should yield accurate results. As well it would display player activity average weekly.