The thing about ratings systems is that, assuming game results are stored in a sensible digital format, they should be pretty straightforward to replace. If an Elo-based system is set up and then in a year it looked like Glicko or something some Netrunner-playing math wiz came up with would be better, it’s possible to just run the new algorithm on the historical data and announce that everyone’s ratings have been converted to the new system. Changing the system every year would be a bad idea, obviously, but it’s definitely possible to recover if the first choice turns out to have problems.
The trickier bits to decide, I think, are things like: what games should affect ratings? How will game results be reported? How will disputed results be handled? How will players be tracked to make sure results are associated with the right player? If results are published online under people’s real names, how will players who do not want their real name published be handled?
Here are some possible answers to the above, for people to argue with: count games from any non-online tournament run according to current tournament rules where the TO signs up to submit ratings; the TO submits ratings; require that paper match slips be used, and require the TO to keep them for a week in case of disputes; maybe some kind of player number, I guess; allow people to request their name be hidden when they sign up for a player number.