Many thanks to those who have looked in to webdiplomacy's usage statistics over the past week or two. It inspired me to go and do some digging of my own. Thanks are also due to HR and PW who helped with discussion and did the imgur legwork.
I've done this analysis directly against the games database. This has the advantage over the GR data of it not being affected by changes in the way GR is reported or player bans, and is a cleaner signal than the Google Analytics data because it's a direct measurement of the games that are being played.
You can see the plots here:
http://imgur.com/a/a2aNL
General conclusions:
* Numbers are down on previous years, but it's not doom and gloom yet.
* The site reliability issues last year *really* hurt us, and I don't think we've recovered
* I don't think the removal of PPSC hurt us, and SoS appears to be a good replacement
* Live games are way down, and that's responsible for the majority of the drop in games
Going through the plots:
Games completed over time: Here you can see a quarterly plot of the number of games finished. There is a small downward trend.
Live & Non-Live Games completed over time: This shows that the majority of the downward trend is due to fewer live games completing. This may be due to a culture shift towards cancelled games instead of draws in live games when there are many NMRs (since cancelled games are removed from the database, they can't be shown here).
Games of each press type completed over time: Not much to say about this one - full press games are the majority of our games, and games in general are down. It shows that rulebook press games haven't really hit their stable number yet, but also that they're not super popular anyway.
Games of each scoring system completed over time: This shows that PPSC was dropping in popularity since WTA/DSS was introduced as the default in 2014. It also shows that SoS has nearly completely replaced it, although the trend is (possibly?) still downward.
--- distinct players ---
Another way to look at this data is the number of distinct players completing a game (this is similar to the GR analysis that Oct did).
Distinct Players over time: This shows that we have a reasonably clear seasonal trend to the number of unique players. Our numbers had been fairly consistent up until the server issues we had around September/October last year, when we reached a new low of 979 players finishing a game in September. This trough is also low this year, at 894. For comparison, the numbers for September 2013 and 2014 were 1,212 and 1,274 respectively.
Distinct Players in Live & Non-Live Games over time: I don't think this one is particularly interesting - it's a similar picture to the number of live games plot.
Distinct Players in different scoring modes over time: similarly to the number of games plot, this shows that there's a clear seasonal trend in WTA/DSS games, and that PPSC is well replaced by SoS/unranked in terms of number of users.
---Bonus plot ---
http://puu.sh/ruPxe/708d6e52ca.jpg
This is number of games in each variant over time. Clearly there's an impact on classic from the variants.
I think this plot shows the danger of adding too many features - because it's harder to get games off the ground if players can't agree on the variant/game options. There's a clear impact on Classic, and I don't think it recovered. HR thinks that perhaps the variants were added to combat a general downward trend, and that's what's visible in the classic plot, minus the variant games. You be the judge.
To quote PW: "It could be that splintering the community results in fewer total games, or it could be that there are fewer games because there are fewer players, unrelated to splintering. Or there could be fewer players because their games don't start because of splintering".
HR pointed out that classic players may be more likely to leave, because there are simply more options available if you want to play Classic.
Either way, it shows that we're not losing players at a devastating rate, although it's clear that some improvements to player retention are necessary.
Modernising the site is a really important goal, especially with the global trend towards internet access on mobile. Additionally, improving the look of the site could help with player retention. I don't think it helps us to look like a dark corner of the internet (although I know many oldtimers appreciate the nostalgic look here).
It's possible that removing some variants could improve the number of games that get off the ground. I imagine this would be unpopular with some.
Having said that, adding some smaller variants might improve player engagement. There's been some talk of adding one or two 1v1 variants - I'd like to maintain the focus on classic diplomacy (so that vDip can be the variant place) - but the Classic dip 1v1s (eg France vs Austria or Germany vs Italy) might be interesting.
This also shows that it might be worth investing some effort into improving the live game experience. Live updating boards would be *amazing*, and we have several plans for better NMR protection. I don't think any site currently does live updating boards, and I think NMR protection is generally poor across the board.
Of course, modernising the site in general is a huge amount of development effort. Last time I spent my summer developing for webdip, I got called Hitler. This summer, I'm going to be on a beach in Vanuatu with no internet access. I don't have enough data yet to tell you if the two are related.
:)
--- some plots that didn't make it into this one ---
From a quick investigation, there doesn't appear to be a clear story for player retention. No one time in the site's history was the best for player retention, and no one time in the sites history was the worst for player exits.
There's also no clear correlation between webdip points and number of games played.
However, users who joined in 2010 play more games on average than any other group, with a mean of 8.2 games (compared to a 6.0 mean overall).
The median number of games for members who joined in all years other than 2006 is 1, though.
Remember, as a great man once said: "We're not sinking; my end is 20 feet in the air!"
Discussion welcome and encouraged.