Age :41
Group: Administrator
Location: United Kingdom
No presentation has been written.
EQL  /  6 Dec 2010, 16:12
Unplayable Lag
Well following the infamous pingwhine debate in the recent TVS-Fusion semifinal game of EQL12 ( http://eql.quakeworld.nu/eql12/matches.php?id=253#Comments ), and subsequent criticism of the rules, I have decided to draft some prospective rules on server selection and pings. Please note that these rules are just that, a draft, and one that needs feedback from you, the community. There are some elements I am not particularly happy with, in particular the ‘65ms cap rule’, and it definitely needs some work in terms of addressing situations which I have not considered. Feel free to conjure up any kind of outlandish scenarios (in relation to pings) and lets work out the best way to handle those in terms of ensuring that they are covered in a ‘fair’ or at least ‘sensible’ way by the rules.

Finally, it should be noted that I don't have the authority to implement any rules, as I no longer hold an admin position in any 4on4 tournaments. The idea is formulate a suitable ruleset. incorporating feedback from the community as deemed appropriate, and suggest that tournaments adopt these rules as their disgression.
The first thing to do, is to state what the current EQL rules are in this regard:

***********

Server and minping
• Choice of server should be suitable for both teams
• If the teams can't agree on a server, clientside minping can be used (/cl_delay_packet xx)
• The value for the minping must not be higher than the lowest pinging player in the highest pinging team, unless both teams agree otherwise. The maximum minping that can be required is 51.
• In cases of American clans meeting Europeans the games should be played on servers where Americans ping the best in Europe. Which are usually UK and NL servers.
• Ping is still considered the decider of the server. Antilag differs on servers and can't be required. Ping is still the last decider but we recommend playing on servers with antilag ON.
If the team with lower ping wants, they can reroute via proxy to raise their ping.

**********

So, what are the flaws in these rules? In order to craft better ones, we first need to understand the problems with the existing ones:

1) There is no clearly defined system for choosing the server, only that “ping is the decider”. One assumes that we are trying to get ‘fair pings’, maybe even comparable averages, but as we all know, to “ASSUME” makes an “ASS” out of “U and “ME”! It needs clearer definition.

2) Setting a minping value the lowest of that in the highest pinging team creates a potentially nasty situation for clans of players with widely different pings. Take CCCP, for example. In Denmark Moltas has 13ms, and the rest 39ms+. Going by these rules, a Danish server would be deemed ‘fair’ even against a team of 4x 13ms. What this effectively means, is that it gives such clans no option but to apply game theory and increase the pings of their lower pinging players, in order to force up the minping required from the other team (affecting up to 4 enemy players but maybe only one from your team). In this case, to gain optimal conditions Moltas should raise his ping to 39ms, thus forcing the opponents to minping 39ms as well. This will raise CCCP average ping by only 26/4 = 6.5ms, whereas the enemy clan has their average ping raised by 26ms – a bit of a no-brainer.

3) Extending this game theory approach, it is inevitable that in such cases you should just ping up to the ping of your highest pinging player because then you can never be at a ping disadvantage (assuming it’s no higher than 51). Why is this bad? Well, because it is forcing some players to play with higher ping than they necessarily need to, in terms of still achieving a fair game. Let that clan keep some players at lower pings and then you get to keep yours down too. Take the tVS-F game, in the end they played with 2x26 2x39 which seemed the best solution to me – only one of tVS needed to ping up to 39 instead of 3. Surely that is better than 4x39 vs 4x39?

4) The rule about American clans playing in Europe is slightly flawed in that it is based on the assumption (a reasonable one, but still an assumption) that the fairest conditions can be achieved by ensuring that the Americans never have to play anywhere but their lowest pinging EU server. This may not necessarily be fair; for example some years ago there was a QW server in Iceland where some of the NA players pinged the best (e.g. 77ms for Bronx, lower than in the UK), yet I’d imagine players from eastern Russian might really struggle to play there. Additionally, “American clan” maybe needs a bit more clarity in definition – what about Canadian players, or predominantly American clans featuring EU players (e.g. O clan)?

And so here is my initial draft of potential rules.

1) Mutual agreement
First of all, try to agree server and ping conditions with your opponent without resorting to the below steps, as this should be much faster. In the majority of cases, mutually acceptable conditions can be agreed without consulting the rulebook. If you think server selection might be an issue for an upcoming fixture, you are advised to contact your opponent before the scheduled start time and discuss your options.

2) Server selection
When there is a dispute over server selection, the primary goal is to minimise the average ping of the higher pinging team. Example:

Server 1: Team A average = 32ms, Team B average = 39ms
Server 2: Team A average = 35ms, Team B average = 26ms

In this case, Server 2 should be chosen because the average ping of the higher pinging team is lowest (35ms vs 39ms for Server 1).

In the event that multiple servers have the same highest average ping, the server where the ping of the highest pinging player is lowest should be chosen. If there are still multiple candidates, the server where there is the least difference between the average pings of the two clans should be chosen. If there is still more than one potential server, and no agreement can be reached, randomise the choice (cmd rnd server1 server2…. serverN)

3) Showing your hand
During this initial server selection phase, all 8 players for the upcoming map are required to ‘show their best pings’ (for this reason you should decide the 1st map to be played before deciding on server). This means that you should use the best proxy route available (“cl_findroutes 1” and “sb_refresh” in ezquake may help, although should not be relied on 100%) and attempt proxies suggested by your opponent who may be more familiar with the routing to a given server. All players should use cl_physfps 77+ (or 0) and cl_delaypacket 0 during this phase unless otherwise agreed.

4) Packetloss
Servers giving packetloss (>2% average) can be rejected if another (PL free) server is available where the average ping of the higher pinging team is at most 13ms higher than that on the plossy server. If the packetloss affects less than half the players, in the first instance it is recommended that proxy re-routing is attempted.

5) Lineup changes
Ideally all maps should be played on the same server; however, due to lineup changes between maps you may need to adjust pings and/or switch servers between maps in order to minimise the average pings.

6) Adjusting pings
Having chosen the server, the next step is to raise the average ping of the lower pinging clan (Team B in above example) so that it matches that of the other team. This should be done initially by recursively increasing the ping of the lowest pinging player(s) on Team B by 13ms. Due to fluctuating pings, and the fact that with cl_earlypackets scoreboard pings are only ‘bandings’ anyway, a difference of 1ms on scoreboard average ping is considered acceptable.

7) The 65ms cap rule
NOTE: For the purposes of calculating the average of the highest pinging team at the ping-equating stage, you should cap the pings of their players at 65ms. So for example, for the purposes of team average, a player with 116ms is considered to have a ping of 65ms. This means that you may need to manually calculate the average ping instead of using the scoreboard. So, for example 26+26+26+116 has a ‘real average’ of 48.5ms, but a ‘ping-equating average’ of 36.75. The reason for this rule is to avoid situations where teams can deploy one very high pinging player to bias their average upwards and thus force their opponents to use an excessively high average. 65ms is an arbitrary value but has been chosen based on the conditions that the majority of European players will face. There is also a knock-on effect that means no player should ever have to ping up higher than 65ms.

8) Quick summary:

1) Choose server by minimising average ping for highest pinging clan
2) Lower pinging clan pings up their lowest pingers to equate average
3) Re-evaluate pings after lineup changes between maps

Any there we have it. What needs changing, what needs adding (in particular with respect to clans from outside Europe).
Comments
2010-12-06, 18:59
2) The rule should be applied to the lowest ping that the player gets to the server. Rising ping artificially to force opponent to ping up shouldn't be allowed. If there is a player on both teams that can get 13ms to the server, then no ping ups are needed.
2010-12-06, 23:28
The problem with prohibiting raising ping artificially is that under the current ruleset it can put clans with a wide range of pings (typically those with physically dispersed players) at a disadvantage in terms of their low pinger being a freepass for the opposition to suggest favourable conditions for themselves. If you think about it, normally the only reason clans would ever need to raise pings artifically is in cases where their clan is at a disadvantage. I just don't like the whole "minping is the lowest ping of other team" because it potentially sets the pings of 4 players based on the absolute ping of 1 opposing player - at least when dealing in averages it dilutes the effect of extremes in pings.

Of course some might suggest that we should, in fact, actively discourage clans from having wide ranges of pings for the good of the community - bunch everyone together and avoid the need for some much excessive use of minping. The problem is that clans featuring players of only one nationality is now the minority, rather than the majority. Taking div1 playoffs as an example, of the 8 clans, only QK has players of a single nationality. Even if we widen the parameters a little be players from adjacent nations, 6 of 8 clans still have dispersed players.

Probably the most 'extreme' example I can think of in relation to my rule, would be a team having 13+13+13+65 vs 13+13+13+13. This first team has average ping of 26 and thus requires all the enemy team to ping up to 26ms. This means that 3 of their opponents have lower ping. I'm not sure how popular that would be - maybe there is a better solution, maybe rather than having a 65ms cap something better can be achieved by instead of capping the maximum ping used in the average to an absolute value, base it relative to the average ping of the other players. e.g. how about, the maximum ping of the highest pinging player used for the purpose of average is considered to be capped at the average of the other 3 players, plus say 26ms? Still problematic in itself of course - what happens if more than one HPB is present? Hmmmm....
2010-12-07, 11:08
Good suggestions. However as far as I know the only game with ping whine for the whole season was the F-TVS game? For everyone else the rules seemed to work just fine!
2010-12-07, 12:11
How many of the about 150 games did you spectate live Hagge to come to that conclusion? We had the same kind of trouble in one game for example.

Plus how many games have been played with unfair pings just because the higher pinging clan doesn't think it's worth the drama. Or is just used to the situation and think that they deserve to play with worse pings since they don't live in Sweden/Denmark/Finland.

Rules shouldn't be guidelines, they should be precise instructions that can be followed to avoid the drama.
2010-12-07, 12:34
Ok sorry Medar, must admit that I'm not too familiar with what's been going on this season. I have myself never had any ping issues when playing a game for many many seasons.

The suggestions by HT are very good though and hopefully some new rules can be created with his feedback.
2010-12-07, 14:49
#4 +1

2010-12-07, 18:32
Hagge, the rules may work 'just fine' in the majority of situations, but that doesn't mean they aren't necessary - as I said in #1 'Mutual Agreement', there is nothing to stop clans continuing to play games in their normal fashion by agreeing on server, maybe tweaking something here or there and then playing the game as per usual.

However, while it doesn't matter if you have complex rules and agree a laissez-faire approach to following/applying them, it DOES matter if you have laissez-faire rules and a complex approach to applying them. Rules are intended to be a fallback failsafe option for when there is any sort of dispute. They need to be unambiguous and written in such a way that their application does not allow (or at least minimises the potential for) clans to exploit them for their own gain.

As Medar alluded to, we had one problem this season that was so bad it resulted in players dropping and the game had to be rescheduled, not to mention all the frustration and (perhaps unnecessary) ill-feeling it may have caused.
2010-12-23, 06:52
Valiant attempt at solving an age old problem. I do like the proposed rules. The nature of the internet and this type of competition is that there will never truly be a fair environment. Like many others I am from the old school era where we planted our feet in the ground, stated I am, and played with whatever pings we could get. There is a lot of opportunity in the quakeworld community to stop the whine, grow a pair of Wanderlei Silvas, and gain some respect.
You have to be logged in to be able to post a comment.
Username:
Password: