My research here is not very scientific or exhaustive, since these groups don't publish their results in a uniform way (or at all). In some cases I was forced to dig nuggets of speculative information out of blog posts from former team members, and in every case the numbers are somewhat old.

So take this with a grain of salt.

One last note: since Lines of Code does not necessarily tell the whole story and is certainly not the ideal metric for determining which project took the most effort, I included a few other relevant details that came up.

The Contenders

* Surprise! You have probably never heard of this, but the 5ESS Switch is a giant piece of telephony infrastructure that came out of AT&T. It runs about half of the telephone exchanges in the US. It isn't clear what it was written in or if the 100 million metric is actually reasonable, but a lot of people worked on it for about 20 years.

** This one is tricky. The Linux kernel itself is, as of 2.6.28 (December 2008), not too far over 10 million lines of code. At the same time, some sources say that Fedora 9 is well over 200 million lines of code, but that includes all the various packages that it comes bundled with. I thought these numbers on Red Hat, which come from Wikipedia and a few other sources, looked like it offered the fairest comparison to the Vista numbers.

*** This is only rumor... However, I found corroborating mention from a few different sources, including one Microsoft employee who said that each Office dev owns nearly 500,000 lines of code and that they have extremely high turnover. The fact that Microsoft isn’t trumpeting anything about the size of the Office effort may reinforce this. If it's true, it's fascinating that such a small team could maintain a project of this scope!

Other important Projects

Future Combat Systems

This is the US Army’s ongoing modernization program. It is truly massive one estimated to end up at around 63 million lines of code. It involves perhaps as many as 2,500 developers, but since it’s an ongoing project that won’t be ready for use for at least another 7 years, it doesn’t qualify for the list.

Conclusions

If Wikipedia is correct, the 5ESS Switch is the largest successful software effort in human history. This makes a certain amount of sense if you consider how big (and old) our telephony infrastructure is.

The US Army might eventually wrest the title from them, since Future Combat Systems is supposed to keep going until at least 2030 and, if I know software, will probably grow. That would be a truly enormous project.

As for the battle between Vista and Linux, it may be too close to call. Linux appears to have far more individual developers touching it, but a good chunk of them only contribute enough to say that they were involved and then move on. Whereas Vista’s numbers reflect the size of a full-time team without (to my knowledge) factors such as churn worked in.

On the other hand, Windows seems to have more lines of code, but then there’s the Fedora 9 metric that puts that particular distribution at over 200 million. So it’s impossible to make a clean comparison from the data that's available, and I going to pass on the opportunity to get involved in a religious war by drilling down into a comparison of features and which OS is “best”.

So there you have it. This may be murky, but hopefully I have at least illuminated the scale of these projects a bit. If you happen to know of any other epic projects that should be on the list, I would love to hear about it!

software massive comparison