This is primarily intended for Fizzer, but someone else with insight into this feel free to respond.

Some time ago Fizzer introduced a plan to remove players from the ladder who were inactive and getting booted. I recently noticed on the 1v1 ladder feed a player named Rossen who has a sub-800 rating, so I checked it out. Numerous boots over a several week period, and totally inactive on the site for 23 days.

Are players still automatically removed from the ladder for being inactive? If so, how did Rossen remain on the ladder this long? If not, why not?

Since it's a paid account, Rossen will not have the option to make a new account to clear their record of boots. Being on the ladder while away, they are automatically put into new games. It's ruining their account and also ruining the games of the people they get matched with, who have to wait the six days to have an automatic win against someone with such a low rating that their own rating might even go down.

They're removed. After...quite a long time.

It was the main reason I made Omniscient. My previous account got booted about as much as Rossen has (or perhaps more, not checked how many times Rossen has been booted, but I got booted till auto-withdraw kicked in).

I think the auto-removal should probably have a much faster reaction, since removing the people from the ladder not only benefits them, but also benefits the other people on the ladder since they aren't assigned games against people that won't actually play the game.

The way it used to work was that Fizzer would personally remove people that he figured were not coming back. Often this took about a month of inactivity on their part. He also only seemed to notice the autoboots for good players, who played in the top 10. Bad players losing a lot doesn't garner as much attention, I guess.

Anyway, there was a uservoice petition and a previous thread on this topic, pointing out some people who had been inactive for a very long time. Fizzer responded that he would set up a system to automatically remove inactive players. I belive the timeframe he mentioned was something like 10 days of inactivity. Still enough for someone to get booted, but much better than 30+ days. I believe that this system did work for a period of time, though from our end it's not easy to see if someone voluntarily took themselves off verses the system.

Rossen seems to indicate that either the system was never made automatic, or it was taken offline.

Players are automatically removed from ladders if their account is over 20 days inactivate. And it did automatically remove Rossen, you can see he's been removed: http://warlight.net/LadderTeam?LadderTeamID=1696

Have you considered lowering that timeframe down some? Or could you at least explain why it is so high? Someone with five games at a time could get booted from more than 20 games in that time period, and I can't think of a single benefit to having it be that long. There's certainly no benefits from people getting booted or even people winning games to booted opponents.

If you're worried about making decisions on behalf of people, that seems like a very minor one to worry about. If they intend to play in the future, most people would be happy to avoid more boots. If they aren't coming back, then you can remove them on day 1 and do everyone a service with no harm at all. You could have a generated e-mail from the system that informs the player that they have been removed from the ladder due to inactivity. I doubt anyone would have a problem with that, and it lets them know that they need to re-enter the ladder to get more games.

I've said before that my preferred solution is six days, which is enough time for new games to go all the way to autoboot. Even ten days would be much better than twenty, in every way.

Right now, the only thing the algorithm looks at is how long the account has been inactive. If the inactive threshold was set too low, there are conceivable situations where someone could be active but still be kicked from the ladder when they shouldn't be.

Say you set it to 10 days. Now presume that someone's in the 1v1 ladder playing two games at a time and not playing any games other than the ladder. Now presume both of their opponents activate a 10-day vacation. In this situation, they'd be kicked from the ladder, which is totally the wrong thing to do since they were fully active and just waiting on their opponents.

The algorithm could certainly be improved by looking at other metrics, such as if they've gotten booted from games recently. But right now it's very simple.

Honestly I don't think it's as big of a deal as you think it is. Their losses will expire, and their boots are deserved since they didn't leave the ladder. Assuming they aren't making a habit out of joining a ladder and abandoning it, their boot count on their profile will be insignificant compared to the number of games most of us play.

It does take up a slot for someone for 6 days, but if they want more games they can always just up their game count by one.

And I don't buy the argument that beating someone below you lowers your rating. I've ran simulations before where the #1 player beats the very lowest player on the ladder and it didn't reduce their rating at all. If you can prove this statement via running a simulation, let me know. (instructions are on the wiki).

I agree. I think 20 days is plenty low enough.

If there is an inactive issue that needs more attention, it is in tournaments without an auto boot. Right now it takes 100 days for a game to go away if neither player shows up and there is no auto boot. I think that needs to be lowered a lot.

I did not consider the possibility of a person's only opponents all being on vacation, though I think that's a pretty small chance. I'll agree that having a 20 day removal is a lot better than no removal, and that it's not a really big deal anymore (it used to be a much bigger issue). I guess I just hate to see people get booted at all, especially when games are automatically added.

About losing rating due to a win, that's something I've heard a couple of times from good players and had not seen refuted. If the math doesn't work out that way, then that's good.