according to fizzer the boot percentage already got lower. https://www.warzone.com/blog/ but I dont trust a statistic with 2 values.

Edited 6/1/2023 09:46:51

^ As Dan pointed out in another thread, it's inconclusive, and could be 1) because of the influx of mobile players due to ladders being on app now (and maybe people elsewhere who now see it in the "multiplayer" tab) and 2) because players who have doubts whether they can keep up simply leave rather than play and risk getting booted

Ah, I just read the blog - boots decreased, but I guess that just means that the "remove player from ladder after he boots" functionality works.

@linberson

it worked so well, it kicked you for using -Auto Commit-

I'll compare participation and how often players boot for each seasonal ladder season so far. Can't use 1v1 ladder as a good comparison because of being limited by last 30 days in the API.

You can't compare it because multiple measures got implemented at once.
It's impossible like this to find any real effect of each separate measure.

The only thing that you can measure, and that Fizzer is interested in anyway, is QoL (quality of life), i.e. if players are happy with this.
That too is near impossible to measure in practice because people will always inadvertently directly compare with the past ("better" or "worst") versus their quality of life in either system at the same point in time, independently of what was and what will be.

If we are talking about quality of life, for me personally, it's very easy. I'm wasting both my time and mobile data while on holidays, realizing there is likely no more future in this game for me.
I have only one reason left to play at this moment: I was not aware of the change before the start of the seasonal and got invited to a bunch of seasonal games (with the changed boot times), and I would hate to get another one of my seasonals ruined by boots (my only not top 100 ranked was due to boots). After this seasonal, I'll need to have a long and hard discussion with myself if it's worthwhile to get stressed over an online game for the sake of keeping my last streak (not ever missing a seasonal) alive. I fear I already know the answer.

^ so participation would be reduced either way because the boot times are far too short? That's the point i want to discuss rather than players getting annoyed by others taking 'too long' to take their turn without deliberately artificially effecting their ratings.

There's clearly multiple factors to consider either way because each season should have different settings to others. I'll get the data but not any conclusions. Only saying what the data suggests

Edited 6/2/2023 20:06:02

To go in a little more technical (statistical) detail.

In essence, what you want, is to tease out the effect of each of the different measures that were taken on one or more outcomes. Each outcome that is chosen in imperfect, because it only tells part of the story.
However, each comparison is majorly flawed because any difference can be cause by a multitude of factors*, both at the participant level and because all the measures were rolled out at once. In statistical jargon, this is called "complete confounding"**. When you have complete confounding, it's impossible to find the effect of each of the factors that are confounded. In addition to the complete confounding, you also have several other (partial) confounders, which would need a careful study design and analysis to tease out. None of that is done here.

Therefore, I can tell you, well in advance, that it will not be possible to correctly translate whatever you see in the data to the real causal effect. Everyone will be able to fill it in based on their predisposition ("I hypothesize", "I believe", "It seems most likely to me") and anecdotal evidence (i.e. "someone said", "I heard", "For me personally the reason").

* Known variables and confounders in this case (and there are probably a bunch more so-called "unknown confounders" and a few that I didn't think about)
- boot timer
- booted people immediately removed
- new rating system rollout
- template change (I heard?)
- template difference (seasonal)
- time of the year (e.g. exams in many countries)
- (potential) boycott by some players / public opinion
- different season length (seasonal)
- changing ladder joining behavior
To make it very clear, even a list of 3 or 4 would be potentially be extremely problematic when not collaborating with an expert statistician from the start to help you with design, modelling and interpretation of a study. Here there are likely 10 or more.
Many of these have a massive impact on which players will join the ladders and when. Most of them aren't even measurable...
As a former examinator, I can tell you, the best way to increase the average grade of your students at the exam, is not by changing the exam, it's by murdering a significant portion of the weaker students. This is in essence what may be happening here: improve the results by eliminating a bunch of undesirable players from the ladders, directly and very much indirectly.

**(Example from my former statistics professor: a PhD student who compared yield of different fertilizers, but used them in different years. The effect of year and fertilizer is completely confounded, you can not measure fertilizer effect because any difference you measure may be due to differences between year, e.g. amount of rainfall and sunshine, and other factors directly influencing yield such as quality of the soil varying from year to year)

So based on that everything is inconclusive? That’s good enough for me. The data doesn’t lie but has definitely been misinterpreted by Fizzer.

Should I really be looking for if lots of players suddenly booted during this season at the same date?

Nobody talking about how # of booted is rising from week 17 to week 18?

That's probably to be expected because of getting a larger player base on ladders from app. It's hard to tell without getting all the data and seeing which new players have joined

So based on that everything is inconclusive? That’s good enough for me. The data doesn’t lie but has definitely been misinterpreted by Fizzer.

In a sense, yes. While the data may not be inconclusive, and may or may not show a strong effect, everyone can just make their own narrative about that: "It's because of the boot limit change" vs. "It's because booted people get removed" vs. "It's because of people being more enthusiastic" vs. "It's because of the new rating system" vs. "It's because of the people playing this template being more used to lower boot times" vs. "It's because it was bad weather outside so most people had more time to play" etc.

None of that could be proven nor disproven with the way that the data is structured, so you risk that everyone just sticks to their entrenched opinion no matter what. Very often, collecting and analysing data is useful, but there are times, sadly like this, that any data-analysis is dead in the water before even starting it.

Has Fizzer misinterpreted the data? Very likely, if this was an analysis in a scientific paper and I was the (statistical) reviewer, I would reject it immediately. But we likely cannot proof that he misinterpreted it either.

Technical note: some things may be possible, but it would require an extensive data collection process, including getting the reasons for boots before and after, modelling if people would have booted again if they weren't removed and/or discarding boots of people who would have been removed, and correcting for a bunch of other variables to the extend possible, for example through matching or propensity score matching. The time and effort required versus the small chance of being successful and the limited impact that success could have should disqualify such a study immediately.
However, it's your time that you invest, I'm just trying to warn everyone here to not expect too much from it in this case.

Can't do a full comparison, over 30 seasons dont have any games.

There are 52 seasons listed on the season ladder page, so I assume some IDs have simply been skipped.

If you got 52 JSON files, you got them all :)

Maybe I should have booted in all my ladder games rather than just surrendering because I hated playing it everyday.

Let's all enter ladders and boot to increase it.

Edited 6/6/2023 20:42:57

+1

@Dan, here are the ladder IDs if needed. I assume they should all work with the API

4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4065
4066
4067
4068
4069
4070
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086

Edited 6/7/2023 14:59:04

4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4030
4086
4029
4031
4032
4065
4066
4067
4068
4069
4070
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085

Thanks! That's a very nice find.