Tuesday, May 8, 2012

Garbage in, garbage out

PC Gamer - Complete History of Gaming
PC Gamer - Complete History of Gaming (Photo credit: Defragged)
When you want to prove something, you have to prove it with data, not bogus extrapolations.

There was an article on Gamersblog back on November 2011, that claims PC games, due to digital distribution like Steam and such, is going to destroy the consoles. Their "proof"?

Battlefield 3 PC outsold Battlefield 3 PS3, close to BF3 XB360

Wait, where did they get *that* idea? They said they "extrapolated" it. Through these steps

1) VGChartz got the figures:

Xbox 360: 2.2 mil­lion
PlaySta­tion 3: 1.5 mil­lion
PC: 500,000

As of October 2011.

2) They "estimated" that 75% of sales are made digitally, which is NOT counted by VGChartz (who only counts retail), so the "real" PC sales is actually 2 million

3) They "estimated" 75% because "Accord­ing to Techcrunch as of 2010 it was esti­mated that over 57 per­cent of PC gam­ing sales are done dig­i­tally and that num­ber has only grown"

Sounds reasonable, but what does the Techcrunch article actually say, as they didn't link to it? Well, I found it.


It actually says: "According to the all-seeing, all-knowing NPD Group, there were 11.7 million digital sales of PC games during the first half of 2010. That compares to 8.2 million physical sales. "

First of all, using early 2010 data to extrapolate sales in late 2011 is pretty bogus. furthermore, BF3 is only sold on EA's Origin, which doesn't exist until June 2011. Which makes it even more bogus.

Second, that survey is for all games, casual (like those hidden object games, or simple games like Drop, Zuma, bejeweled...) AND big AAA titles. And this is TOTAL SALES. You can't use this to predict ONE game's distribution of digital vs. physical sales!

To give an example: 100 vehicles are sold at dealership A, and 80 vehicles are sold at dealership B. Five vans were sold at Dealership B. How many vans are sold at delaership A?

The answer is: you don't know. You have to ASSUME that both of them are equally adept in selling vans (thus, the van to vehicles sold ratio is same at both dealerships) to get an answer.

Same with the game sale stats. You don't know what is the distribution of sales is in both sales mediums (retail vs. direct online). So you can't use that to extrapolate!

Third, TechCrunch didn't do any predictions, but actually linked to its source: a Gamasutra article and what does THAT say?

"... the tracking firm [NPD] appears to lack agreements with the majority of digital retailers. This means that the figures cited are extrapolated from third-party tracking. The firm attempts to divine PC game purchase habits from surveys, using anywhere from 8,000 to 180,000 individuals on the NPD online consumer panel, depending on which of its methods it chooses."

In other words, the data is NOT reliable, but are very rough estimates! They have a large margin of error! And it does NOT predict any trends and such.

When you use it to predict trends, or to "prove" a trend, you get GARBAGE. Garbage in, garbage out.

Which was borne out when EA released sales figures recently...  It says that Origin, during 10 months of existence, pulled in 150 million.

Consider the fact that catalog of EA PC games sold by origins includes the SIMs (all permutations), and many other games (it numbers 202, as of May 2012, so should be over 100 back in October / November 2011)

Consider the fact that EA serves ALL consoles (Wii, PSP, PS3, PS2, NDS, Xbox360) on Origin, in addition to the PC

Given that the retail price of BF3 on Origin is 59.95, to sell 1.5 million copies of BF3 on Origin (75% of sales ought to be digital) would account for... 90 million of that 150 million revenue, or fully 60% of ALL games sold by Origin.

Clearly, it is NOT possible. Thus the extrapolation is wrong. And the premise is NOT proven.

Enhanced by Zemanta

No comments: