So, I have two FOSS "origin" stories, if you will, sort of my counterparts to the (in)famous frustrations with a printer driver about which you may have heard. It's how I remember coming into this FOSS world.
License frustration
In the first, I went to go use this suite of molecular modelling software, so far as I know the only copy on campus, called Biosym Insight/Discover. I don't know exactly the name of the module I was using but I found a screenshot that shows some things that would have been relevant to me at the time, the "builder" and "biopolymer" options in this menu
The department I was in had a mixture of Suns and IRIX boxes. I think this ran on an IRIX box. Anyway, there was only this one copy, and I had to go into another building to use it. There may have been some fussing around with getting the key, I don't remember. In any case, I go over to use it only to find that I'm not allowed, because our license wasn't considered paid-up and current by the software installation. This drove my interest in other things I could use, and when eventually I found not only were there other things I could use on our lab's own IRIX box right there in the same building, down the hall from my office, but that there were things I could run on Linux at home, there was no looking back.
Hardware taming
The second one was more practical and logistical. We had two instruments in the lab called HPLCs (an initialism for, depending on who you talk to, high pressure liquid chromatography or high-performance liquid chromatography). You could think of these, in loose terms at least, as really, really thick filters. You put your mixture to be separated on the top, and then force it through using lots and lots of solvent. Some stuff would move through it more quickly than others, and you could collect what comes off and record how much and when, and this would give you a good idea about what was in the mixture you put on. If you thought you were putting something pure on, you could tell how pure.
Anyway, we used this thing a lot. The information it recorded was in the form of XY plots, with the X axis showing how much time had passed since you put your stuff on, and the Y axis showed how much stuff came out at any given time. Most of these plots were important research records to keep. In the old days this data got plotted on plotter paper with pens, but by the time I got into the business it was recorded electronically on a computer.
The only snag was that this computer was running the IBM operating system OS/2. At the time, I didn't know that much about OS/2's ill-fated history as a joint venture between IBM and Microsoft. I did know that the computer we had only support Zip disks and we were filling those up fast. Also, the computer had a network card over which we might transfer the data to a system that had better storage (a CD-burner) but the version of OS/2 we had didn't support that. It might have come with an upgrade to the OS, but would that upgrade of the OS still be compatible with the instrument software? We really couldn't afford to screw around with it and possibly put the thing out of commission.
What I did discover was that I could use some combination of root/boot floppy pairs and UMSDOS on the Zip drive as a root filesystem to run the machine from Linux, and then run the network card under Linux. This allowed transferring data over the network to the IRIX machine to which we had attached a SCSI CD-burner. CDs at 670MB for a dollar or two apiece beat ZIP disks for 100MB or even 250MB at 10 or 20 dollars apiece, easily. The burner was more expensive than the ZIP drive, but the marginal cost savings paid that off pretty quickly.