John Maddog Hall has some interesting examples where FLOSS enhanced return on investment for business. It’s interesting reading.
I think ROI is an important concept in education. Since we can get everything we need from FLOSS in education one could put a value on the software similar to what non-free/commercial software would cost and divide by what it costs to implement. I can apt-get install OpenOffice.org for every GNU/Linux box in the place in a few minutes and get the value of eighty times the cost of Office, around $4000. My time for, say, 5 minutes of work is only a few dollars so the return on investment is 1000:1 and we have broken even in minutes.
The entire image we have installed on our new PCs took me a long weekend to polish, say $1000 worth of work. What would be the price of software licences for 12 PCs running that other OS and Office? Probably about the same, but we also have tons of other applications, superior performance and less maintenance. I love having APT able to manage all the apps as well as the OS on a large number of systems. If I have to re-image all the updates from M$ and other suppliers every month the lifetime investment in that other OS keeps growing while I may never have to re-image a GNU/Linux system for years. I can dist-upgrade all units through a local proxy indefinitely. So, I invest $1000 plus a few minutes per day perhaps for years with GNU/Linux and $1000 every month with that other OS. I get $60K worth of value over five years for only a few thousand dollars of effort on my part. I like that, 10:1 ROI over five years.
This weekend I did a bit of investing in our GNU/Linux system that will pay handsomly over the years. When I came here there was no router so I threw a DLINK SOHO router into the breech and it has worked well but has limited features. Today we put DHCP on a server that has proven very reliable and fast so we have now a local DNS with dynamic updates and the ability to point every client at it to cache lookups. I can access every machine by name, too. The result should save a second on every click. What is that worth? Every machine in the building using DHCP will get on-line a second or two sooner. What are all those seconds saved per-user-per-day forever compared to the few seconds it took me to implement it with FLOSS? Another 1000:1 ROI, IMHO, and we should get better performance. Fewer long pauses. Priceless.
I have measured typical DNS lookups coming from the cache at around 5 milliseconds. From the ISP, 600 milliseconds. DHCP from the DLINK router was about 3s often (it runs GNU/Linux too…).