rough cost of wholesale energy per kilowatt hour (kwh): ~5c
CO2 cost per kwh: ~1kg (coal power: http://cdiac.ornl.gov/pns/faq.html)
time for my (small) 1 litre (~ 1kw) kettle to boil when full is ~ 5 minutes which compares well with the theoretical energy for a 1litre at ~350kj, or 350 seconds time for 1kw . Hence power for a small boiled kettle is a killowatt for 1/10 of an hour, or 0.1 kwh

So I get…
Kettle boiling: costs ~.5c, and ~ 100g, … the article says a kettle take 15g, which I don’t get even close to; maybe clever people boil just enough to make single cups only?

If the article was true, Google doing “more than 200m” searches a day would spend ~ $20m a day on power, or ~ $7billion a year, consuming 100,000 megawatt hours, or a continuous drain of 4,000 megawatts (about half the power output of Victoria). On the authors figures, total power consumption would be ~ 650 megawatts, which I guess is about the size of Hazelwood, and would still be spending ~ $1billion a year.

Google use cheap, mass produced low power units in gigantic numbers - estimates are hard to come by, I will estimate 200,000 (http://arnab.org/blog/how-many-computers-does-google-have). Energy cost of networking is significant, but I do not believe as great as machines; I’ll add 50% for good luck. Utility server machines are dropping in power (~100-200w) but also require cooling, UPSs and network etc., so we’ll call it 500w all up (figures are difficult to get; everyone is selling something power center wise) - so I get 100 megawatts; or 1/6th of the author’s estimate, or 1/40th of the true kettle figure.

I’d say that the author is overstating the case to make a political point - and I’d have to agree with the main article it might have something to do with promoting his business…

http://www.theregister.co.uk/2009/01/12/google_kettle_green_it_cobblers/