In the Australian Computerworld article about there being not enough developers of OOo/Star Office, Ken Foskey is quoted as saying there are now 50 people working on the program for Sun in Germany, ten for Novell, and only four independent developers. So much for the magic of Open Source.
Erwin Tenhumberg replies that he won’t talk about numbers, but that the ratio is misleading. This caught my eye because three or four years ago, when I first started using the program, one of the German developers told me that there were over 100 people employed on it in Hamburg. I can’t run down the email right now, but it did appear one one of the public lists. We have been reading for years about Sun making fresh rounds of layoffs, without any specifics. Occasionally well-known and admired sun developers have vanished from these lists. But to go from around 110 developers to around 50 in four years must have had a bad effect on the project.
UPDATE: three or four Sun developers piled in on the lists to say that there were still plenty of them, and there had never been more than 50 developers in Hamburg at any time. The rest were QA people and other support trooops, if that’s the analogy.
I’m not so sure. It depends much more on the qualities of the people making up the project than the quantity.
I also have a suspicion (that I can’t back up in any material way) that there is an optimal size for engineering or software teams which allows good ideas to get implemented quickly and without too much fuss – provided that they are well managed. I’d go for a figure of 20 to 50 people. Above that and things always seem to get too clunky for my liking.
Well, there’s at least one guy working for Red Hat too and quite possibly a fiar few more in their desktop team these days.
On one level you are absolutely right. On the other you are talking errant nonsense.
The fact is that these sorts of projects don’t actually warrant huge numbers of core programmers. The larger projects get, the more communication overhead gets in the way. The fact is that an office product is just not that complicated – certainly not like an OS where there are hundreds of individual bits that can be worked on in isolation – and suffer when there isn’t a single overarching design. Look around at the various small companies that have over the years created replacement office products – Gobe is a good example – all have had a very small core of people – often as small as two or three.
Which brings me to my second point. A good to excellent programmer can, when properly motivated and working on a project that he or she likes, turn out 10-20 times the amount of good code than the average programmer can. Because it is created by a single person it’ll also generally have a better large scale structure than something created by a large team.
I’m a good programmer, I’m have in the past been in teams where I have been responsible for maybe 5 times the amount of good code as the next most productive member of the team. Lest this sound big headed, two of my friends are easily 5 times better than me – possibly much more than that. If I were founding a startup, I’d want someone like one of them to be the core of my product development team.