Archive for the ‘Microsoft’ Category


About FLOSS phenomenon, and the rest…

January 21, 2007

After 15 years of existence of GNU/Linux, can we measure the impact of it and the FLOSS phenomena ?
That’s what EUROPA ENTR Study did in a quite complete Economic impact of open source software on innovation and the competitiveness of the formation and Communication Technologies.
You’ll see a lot of interesting figures and statistics there. You’ll also read Macro-analysis, ie that the most impacted, and active region by FLOSS is Europe, then America, but China and Latin countries will soon overtake Europe.
After leading the field of supercomputing (Top500 Operating system Family), with 75.20% being Linux (17.20% Unix), OSS is gaining popularity in Operating Serrver Systems in companies (15.7% OSS in UK, Sweden, Germany).
While the North/South fracture is disappearing for some country in some domain, the digital fracture is present :
World Map of Commiters by population
Some other interesting questions from the study are :

1. What were the actual costs of developing a given set of software, in terms of time and effort and the equivalent in monetary terms?
2. What is the substitution cost of the same software – i.e. how much would it have cost to build the same software entirely within a single firm in a proprietary software development model33.

Then are given figures about Debian 3.1 FLOSS distribution (2005) :

Source lines of code : 221,351,503
If Debian was written in a software company…
Estimated effort : 163 522 person years
Development cost estimate (till 2005) : Euro 11.9 billion
Development cost estimate (till 2010) : Euro 100 billion
Copyright © 2006 MERIT. Source: URJC estimates (cumulative effort estimation), (salary data).

A note concerning this figure, it uses a COCOMO model adding complexity, without it it would be Person-Years = 2.4 * (KSLOC**1.05)) / 12 = 81915.
Firstly I didn’t know that latest Debian distros were that big, latest stats I read around about the Debian 2.2 potatoe were around 56 SLOC (56 Millions of Source Lines Of Code).
I guess that with those latest counts, X, Mozilla, Desktops are included.
Anyway what are 3 SLOC (for X), 6.2 SLOC (for KDE Snapshot / > 3.5.5), 1 SLOC (KOffice) next to the gigantic 221 SLOC.
You can do your own statistics using sloccount on your favorite projects (I did for X11R6.8.1, KDE Snapshot / > 3.5.5, I also did Beryl 0.1.4 : 0.2 SLOC).
So this has to be considered to the commercial equivalent Software :
Mac OS X 10.4 : 86 SLOC (Most of which must be from BSD…)
Vista : Said to be around 50 SLOC
So what to conclude the more SLOC, the better ? Not really, it’s hard to draw conclusion of the quality of a software only from the number of lines of code, but you can estimate the costs quite easily and you have an idea of how many license one needs to sale to make benefits.
Microsoft problem can be summed up easily :

Suddenly, the market changed and competitors started delivering technology at the speed of the Internet,” said James McQuivey, professor of market research at Boston University. “In some cases, they do it for free, and that’s painful for Microsoft.”

Vista is said to have cost 10 Billions Dollars, it is predicted to be installed on 76 Millions of computers, and should be responsible for $11.5 billion of profits by the end of June 2007.
That’s where we go and what European Study doesn’t emphasize, while there’s no doubt that the FLOSS is growing inside companies, the Personal Computer has a majority of Microsoft OS installed. So why Vista should be the last of it kind ?