I am going to preface this by saying I'm somewhat reluctant to "answer" this as there there is no way this can be definitively answered. We can however make some reasonable assumptions off some known points and look at the two biggest online providers, Lichess and Chess.com.
Figures as of 2023:
- Lichess has ~5,000,000 games a day
- Chess.com ~10,000,000 games a day
To provide these services:
- Lichess has approximately 35 servers
- Chess.com appears to be partially cloud based and scales horizontally.
- however we can semi-reasonably just double the known lichess infrastructure. 70.
Servers similar/same to the ones being utilised by lichess and chess.com generally 1900 - 2100 kwH annually ,
2000 * 105 = 210000kwH/year
Each kwH generates approximately 300grams of CO2 emissions.
so 210,000 * 0.300 = 63,000/year >> 63 Tonnes/year of CO2 emissions.
This is 0.0042
15,000,000 Games a Day >> 172kg of CO2 emissions per Day
0.0115 Grams of CO2 Per Game.
What needs to be understood is that there is a 2 way transaction that is occurring. There is a provider , but generally each game requires 2 users.
We have no way of reasonably calculating the users usage / consumption and kwH , so the best we can reasonably do is use the figures we ball parked from the providers. So for 2 users and a provider
((3*172) * 365.22 ) = 188469 >> 188.5 Tonnes of CO2
188.5 Tonnes for 5.5 Billion Games
Again; this can't be definitively answered. This is only something to get thoughts happening.
Points of Consideration:
- Carbon emissions are based of US emissions per kwH, countries such as Australia that uses coal as a primary power producer, would be higher.
- As previously states this is factoring in the Users. This is going to be the significant factor and at this point in time I cant see an adequate way to tackle this problem without significant data dumps from providers. Generally speaking however, user will use more kwH/game than the providers generally speaking. So these figures could potentially be interpreted as a "starting" point.
- Only looking at Chess.com and Lichess, which doesnt factor in all the other chess providers, the power consumption etc which occurs in OTB tournaments.
-This is only looking at the power estimates per game and not other factors such as general power consumption for lighting heating etc.
- Most larger data centres have cost saving mechanisms for things like power. They generally live in big long buildings with a lot of roof area...ripe for solar panels.
- Most Server Hosting companies provide "eco-power" models where you may pay fractionally more for "green power".
All units are either universal or metric, eg Tonnes is the metric tonne of 1000kg