There are a number of issues that seem a little buried in the EL architecture that I'm hoping are addressed in ES2. If they aren't, it's probably too late to address them now, but I want to put them on the radar for future Amplitude game.



Save games. The handling of save games in ES is, well, terrible. The game you save is often not the game you load. Certain aspects of the game (battle) are not saved at all. It sometimes appears that AI and/or neutrals get an extra turn-- at the very least, an extra think-- upon reloading a game. And certain random characteristics change on each reload, for instance, which heroes are loaded into the market, city ownership penalty decay, and whether winter ends or begins. It's clear that there are some serious issues with how the game handles this, and I think Amplitude would do well to study how it's handled in other games. There are quite a few open source games that do saves perfectly, so there is material for study.



Random seeds. This is kind of linked to save games, in that good seeding of your RNG(s) would alleviate some of your save problems, but it's important for other aspects of the game as well. Consider how useful it would be if you could recreate any game given all of its inputs. You would never have to say, "That's an interesting bug-- do you happen to have a save game from the turn previous?" With good seeding of your RNG and recording every input (including the time of input), you could recreate the entire game, at any point, from any save, with a file that was only fractionally larger.



Automatic error reporting. Amplitude has done some strong work with regards to creating a community interested in improving the game, but there are a lot of errors where community involvement is unnecessary. The number of desync logs reported on these forums is very low. This is a perfect situation for automatic reporting. Your player/tester is already online and is already devoting his or her bandwidth and CPU to the game. If desync logs are useful to you, there is no reason not to create automatic reporting of these. Automatic reporting could be expanded beyond this: crash logs are always useful, and in-game data collection would give you a good eye on balance issues, because you could easily see, for instance, that certain units were ignored by the player base, that certain factions were unpopular or were disproportionately victorious, that sort of thing.



Note that it's very easy to do automatic reporting poorly. Taking too much time in generating a report (it should just be an uploaded log for processing on your server) or requiring confirmation limits the reports you receive.



Client-side prediction. The international nature of Amplitude's audience and the turn-based nature of its games mean that >1000ms latency isn't too uncommon. This shouldn't be a problem, but it is. Unit movements seem to rely on confirmation from the host before beginning, but this is usually unnecessary. In turn-based games, almost all actions can be fully predicted by the client. In the rare cases that the host disagrees, which I think would happen only with timing issues, a pop-up could be generated to draw the player's attention.



Time to display an animation shouldn't depend on the framerate. This is basic stuff that I was surprised to discover wasn't happening in EL. Every single player should see an animation occurring over an identical period, whether that means 120 frames or 1 frame. Yet in EL, players with slow computers are basically barred from MP games because the battles take much longer for them to process.

I agree especially about the automatic error reporting. Oos occurs very often in Endless Legend.

I never even noticed that the durations of animations is dependent on framerate. That really needs to be fixed.

I feel the issues with save games and random seeds are more urgent, though. It drives me nuts at times.