armagetron server search bar..

[Phytotron]

Doing my own "split" from here (although I guess I'm kinda hijacking this thread too, but that one's more specific and unrelated).

And how it and those others got a score of 400 is still beyond me.

Base score 100 + Bonus of 300 for 4 or 5 players + no Malus for version mismatch or >100ms ping. I think the formula is documented... somewhere.

OK, so that explains the formula, but still, does it make sense? I don't think so. If all I care about is how many players are in there, I'll sort by 'Users.' If all I care about is the ping, I'll sort by that. And if I want some optimum combination, I can just sort by one or the other and look. Obviously, at least a partial intent of the score bias is to automate that last bit.

But, how many people actually choose a server based on any of that alone, or even primarily? Question is, should the score bias be something impersonal—it's the server browser's score, not yours—or personal? I think the latter. That's also the most intuitive understanding of it, if one doesn't know what the formula is (and who does, besides the 3-4 of us now discussing it).

So how about ditching that and basing the score on things like how often one selects a server and how long they remain in there. That can be logged, can't it? A bonus for it being bookmarked. Perhaps even a bonus each time a "mate" is located in a server. If there's ever any implementation of some deeper sorting, categorizing, browser tree, whatever, what one prefers there could enter into it as well. Again, just brainstorming, but criteria like that.

And if it's not implied there, the score should be determined cumulatively and stored over time, not just on each load of the browser.

Eh?

[Z-Man]

Yeah, it's definitely an outdated formula. It was from the days where there weren't any different game modes and not even 'high rubber' had been invented and the total number of servers was, umm, about zero and the expected peak server number maybe 10 if I was optimistic. And the single map and single game mode meant that there was a definite fun peak around four players, more than 8 were just crowded. Today, I'd just let the score differentiate between empty servers and servers someone plays on, some servers are fun with two players and some with 18. And yeah, the play history needs recording, you need to be able to sort the list by 'last visited', and the number and duration of your last play sessions > 10 minutes and > 1 match (shorter may or may not mean you tried and hated the server) should be the dominating factor of the score. Servers need to be able to opt out of stat tracking, of course: you don't want an avid tournament player's server list to be filled with tournament servers all of the time.