It reads as if it's from Andy Weir's "The Martian"
Favorite quote from that article:
`Burley adds a closing question: “Does anyone happen to have a 4 mm tape cartridge reader that will work on a modern Linux workstation and a 16-year-old data tape and not disintegrate it?”`
Personally, I find it more interesting to read that older post, and hen follow through the story and updates via comments here and more recent blog posts. Far more interesting to get the original story rather than just a single "up-to-date" bulletin.
This is why someplace where I work there is a computer in a closet that has Windows XP service pack 1, and all the software required to build the code we were working on then. I work in embedded systems, there is always the possibility that we will decide to fix a bug in some old controller that we last sold 12 years ago. I don't think anybody who knows the code is left, but the computers sit, just in case.
NASA should take that to heart and make sure they keep a few old computers in a closed until the last satellite is confirmed de-orbited. Note that this advice only applies to believed dead satellites, satellites that are still in use presumably have a team porting software to something more modern as needed.
Better yet NASA should do what we are attempting: repeatable builds in a container. If I can install a Ubuntu 10 container on whatever the latest linux is and build, I can throw away the Ubuntu 10 computer. This test can be automated with reasonable confidence and low cost.
Look up the crazy stuff that happened with ICE/ISEE-3. NASA had disposed of the high power transmitter that was originally used for communication with it (and other satellites of that era). When it came back around, it fell to civilians to try and reach out to it. The initial contact was made with Arecibo Observatory and a high power amplifier built out of repurposed cellular tower equipment. Fun times :)
I believe what you are describing is a lot of cost with no more benefit for the generally-unlikely scenario of "dead satellite might be recoverable" than what they're doing now: porting software forward on-the-fly. The issue is too rare to have always-running process for at nonzero cost.
Satellites don't really "go" anywhere, generally, when their orbit is known (along short timescales). If porting takes a few days or a month, they'll still know where the target is when ground control is ready to talk to it (especially since they can track it without needing to forward-port the control software).
According this snippet from a Smarter Every Day video it would appear that NASA doesn't port their software to newer hardware during a particular mission (in this case Chandra X-Ray):
https://youtu.be/7AR4yntqLsQ?t=6m41s
> Update, 31 January, 11:20 a.m: [...] Efforts to command IMAGE will likely take another week or two, as the satellite’s old control software is adapted to modern systems.
There's a rule to use original title of an article which sometimes is confusing. But yeah the "Update" bit makes it sound like it just happened or something.
This is quite an outdated post, as NASA has updated as recently as last month. This article that's posted was updated in January. A lot has changed.