No doubt, it it involves quite a bit of work to do it in the first place, and then to get it right, more so if you want the full thing, less if you do a subset at first for evaluation. But I'm not so worried about possible implications of introducing, for opt-in testing, player accounts, and, in a second stage, statistics (which can be presented one or the other way). Generally, we should separate the discussion on the topics involved here: accounts and access control, collection of statistics, presentation of statistics / rankings.
Accounts and access control
While this was mentioned before, I'd like to stress again that I think the account system, as well as collection and presentation of statistics should be opt-in, i.e. you should by no means be required to do any of this. Hosting yourself should always remain possible without any registrations, and if there will be more people providing headless servers then I'm sure some of them will remain open.
It was mentioned that accounts would have to be secure - sure that's always so, and yes, there are cases like Sony and many others where data leaks occurred. Which is bad, and we would need to prevent that. That's even though we are clearly not going to store payment / billing information and hardly any personal information. Which, on the one hand, makes it inattractive to data thieves (i.e. we'd see way less targetted attacks already), and if there was a compromise, the impact would be much more limited, to the possibility of falsifying statistics (which is not much of an issue if you have recent backups which can not be compromised the same way). I' also convinced that some friends would volunteer to do a code review for security.
Collection of statistics
Since it is a matter of privacy, just like accounts, statistic collection should be opt-in only. Some statistics, like hardware / software / network information, could be stored (if the user decides to give it away) anonymously or pseudonomously, so unrelated to accounts. Data related to the players' gaming history would have to be account related, at least in a later stage when it is intended to present player specific statistics online.
Presentation of statistics
On the one hand, there would be anonymous statistics, such as on the hardware / software / network. On the other hand, there could be personal statistics. Those could be visible to just the very player after authentication, or it could be publically available. There could be calculated values, like ranking, or there could be just the values the game delivers already. In the first stage, I think we would clearly not have a public ranking there, yet. This can be an option for later if we realise the foundation of the system (account / permission system and statistics collection) has reached a reliable stage and we are ready and actually still think it's a good idea to have more.
So there are really a lot of options here. We do not need and should not try to have the full-blown thing up in no time, instead we should separate this undertaking into multiple development stages. And we can always decide not to develop or roll out a next step. For a long time, much of this would have to be considered unstable and in-development and a preview anyway, and everyone would need to be aware that any statistics collected may and most likely will be deleted whenever a new preview revision is rolled out. This can and should be well communicated, so that players would not be disappointed about any lost stats.
Yes, all of this is a larger undertaking, but we're not in a hurry, and things can be done step by step, and be tested internally before with some testers, and then rolled out, still as a preview. I think this is doable, and could be fun, and would actually make MegaGlest more attractive to new players (which, despite of the fun of developing it, would be the main reason to do it).
