As the headline points out, a significant activity has been the graduation of our Chess to provide a 3D view of the board. We were aware of the idea of this back in 2008 and contemplated porting parts of our 3D product for the PC to mobile. However then app sizes needed to be small, devices were slow and screens were anyway small, so the idea was not practical. Seeing that more recently a number of Chess programs on Android were providing 3D, we took another look and decided we needed to follow. The first revelation was that in examining what we needed to provide, dynamic rendering of 3D models was not needed. This would only be necessary if the user wanted to rotate the board view, but this was anyway not really helpful. Given the improved support for rendering scaling and rotation by the native Google Play SDK, we found easier solutions. It proved only necessary to pre-render a single version of each piece and then scale it for distance and skew it if it was not in the centre of the view. The latter idea is not obviously viable until you take advantage that Chess pieces are essentially symmetrical. Rotating a rook to the edge essentially means it is tipped over slightly and slightly rotated on the square. Any shadows on the piece will appear to be rotated as well. This might not have worked, but actually it does. The feedback from users has been good.
As part of our relationship with Queen Mary UOL, we took on some work with a researcher Rokas Volkovas, who is working on our Spades program to create an automated tuning system. This is quite difficult given that Spades is heavily influenced by its stochastic nature. You need to sample many hands to compensate for good and bad deals and any such deal needs to be mirrored. We are also working with the very talented Rémi Coulom of Lille University, who had much experience in deep learning and is investigating the possibility of creating a top-level deep learning Chess for us. This would not replace our Chess, but be combined with it.
We made our usual trip to GDC in San Francisco in March, where the number of delegates and presentations is overwhelming. This high concentration of effort delivers a very high quality of presentation and access to networking with the best in the industry. It is a great opportunity to re-align what we are doing with the changing Zeitgeist of the Games Industry. The visit to Google HQ presentations in San Francisco was a reminder just how slick they were. Every moment was planned in detail and I was whisked away to meet people they wanted me to meet. We work with some big entities in the Games Industry, but Google are the top of the pile.
Of course we attended the slick game awards, which is an opportunity to to spend time watching the best out there at the moment.
As ever we re-convened to the R&G restaurant for the annual AI Factory hosted dinner. It is kind of a quiet time when we get the chance to talk to our friends and colleagues, extracted from the mayhem of GDC social life.
Finally we released an updated version of Backgammon with certification by Gaming International Laboratories (GLI) affirming that it had been tested and sources examined, confirming that it did not cheat. This absurdity was necessary as our app has been deluged by reviews claiming it cheats (the same seems to happen to all Backgammon programs on Google Play), which habitually showed the top 3 reviews with 1-star and a rant about it cheating. The impact of certification resulted 62% of reviews claiming we cheat reducing to 42%. Not a perfect result, but better. The Google Play app store has improved the situation by nowadays posting the top positive and top critical review, so users do get to see at least one positive review.
The Summer break saw the CoG (Conference on Games) come to Queen Mary College London. Formerly this was CiG and a desirable event to attend, but usually not in the UK, so not immediately convenient. We attended and had plenty of time to discuss problems we were having with our Spades bidding! Present also at this event was David Silver, the co-creator of AlphaGo for Deep Mind, who gave an excellent talk.