Dotabank began 2 years ago, on September 5th 2013, as a personal exploration into the world of replay analysis, after SkadiStats’ Skadi replay parser was launched. The idea was simple: There is not a lot of game data available via Valve’s WebAPI, but there is a lot of game data in the replays – how much of that can I extract?
I’m not an amazing programmer, so progress was slow, but the core components eventually came together and a barely functional prototype was created. It had graphs of each team’s experience, gold, last hits, and denies over an entire game, as well as hosted a copy of an entire matches combat log (which was HUGE!). It was very slow, expensive, but it had data from the replays – and all you had to do to get that data was stick a match ID in a website.
Then I was hired as a full time software developer. The time I had to spend on Dotabank was greatly reduced, and my motivation for side projects slowly dwindled as the work I was doing sufficiently fulfilled my desire to create. I started to need my spare time to relax.
I realised I would not be able to commit the resources to finish a project like Dotabank. What I had completed so far did not provide useful data for visitors, and I no longer had the time to solve the various challenges that creating a replay-analysis service posed.
I realised something else too. One problem Dota 2 players had, and continue to have, is the fact replays expire. Another problem is there was no service to facilitate downloading replays outside of the game client (which is something I wanted, as my internet at home was very slow). I realised I had built a system that stores player’s replays for them, and without much effort I could add ‘download’ buttons so people could access those replays. Although I was unable to fulfil the original goals of the project, what I did have in pursuit of those original goals was codebase that solved two problems facing the Dota 2 playerbase. So I completely dropped the idea of replay-analysis, and just opted to push Dotabank purely as a replay-storage service.
As people started to use Dotabank, the costs of the service became rather significant – both in time and money. The vast majority of time I spent on the site was maintenance – fixing bugs and cleaning up the (rather inefficient and slow) codebase. There wasn't a lot of evidence of all that work on the front-end; just the site being a little faster, and the costs to me personally – as I fund this project out of my own pocket – being reduced slightly.
With the limited time I had available primarily being spent on maintenance, I lost a lot of passion for the project. Now and then I would gather enough motivation to work on new features (things like browsing replays by tournament, or by hero), and the results of that work were again also minor, but in the end - through the first year of Dotabank’s existence – I simply lost my passion for the project.
I kept the service running however, because people did use it and find value in it. Every now and then I’d receive a small donation from somebody – a token of appreciation for the service which I am very thankful for, and researchers would contact me wanting some of the replays we have stored for various statistical analysis. That made me happy. It didn’t restore my passion for the project, but it at least proved there was some value in the project – and so I have been unwilling to shut the project down. As long as I could cover its costs, I would do so, and that has been my stance on the project over this past year.
With the launch of the Dota 2 Reborn client today, the back-end services I created 2 years ago to download Dota 2 replays ceased to function – they simply don’t work with Reborn. Similarly, the 50,800 replays that are currently archived in Dotabank are now incompatible with Reborn. So starting today Dotabank does not have any replays that can be viewed, and it is unable to download any new Reborn replays.
The problem of downloading Reborn replays can be solved, but at present I have neither time nor motivation to work on a Reborn compatible back-end. The other problem – having 50,800 replays that are effectively useless – cannot be solved, and it is relatively expensive to keep hosting them.
So, after 2 years, 50,800 replays, and 68,682 downloads served, it is time for Dotabank to end. The site is currently in “read-only” mode, which simply the site won’t try to archive any more replays. I will keep the site running for a few more months, so if you want to download any of the replays stored you can, but in a few months I will shut the site down completely.
This may not be the end for Dotabank. I think there is value in the project; not a lot of it, but certainly a little; but right now the simple fact right now is Dotabank doesn’t work and I won’t be able to fix it for a long time, so I’m switching it off.
I am very sorry that I cannot maintain this project any longer, but I am very thankful for everybody who has supported me along the way.
A huge thank-you goes out to those who have donated to the project; I do cover the costs of this service out of my own pocket, and those costs have been a significant chunk of my salary – so any donation has helped me a lot – thank you.
Thank you to those who have gotten in touch and expressed their appreciation for this project, as well as my other work in the Dota 2 space; it feels very good knowing that others have found value in the work I have done.
And thank-you to the friends who have helped me with this project over the years, in one form or another. I’d undoubtedly forget somebody if I attempted to name those friends, so I won’t even try – but those friends know who they are.