We Diagnosed and Fixed the 2023 Voyager 1 Anomaly from 15B Miles Away [video]
youtube.com
youtube.com
asm/disasm can help to find typos in listings, they can help to find xrefs or even to do some static analysis to check for mistake classes they knew they could make. It wouldn't replace any of the manual work they've done, but still it can add some confidence on top of it. Maybe they wouldn't end with priors 50/50 for the success, but with something like 90/10.
Strange. Do I underestimate the complexity of writing an asm and disasm pair?
Writing an assembler for a bespoke CPU is one thing, many of us have done it as a toy project, but stakes are a bit different here. You'd have to mathematically prove your assembler and disassembler are absolutely 100% correct. When your only working model is utterly irreplaceable and irrecoverable upon error, it probably takes a lot more resources to develop.
Which is engineering-speak ridiculous.
To debug anything you need a mental model of the thing in question. With that mental model you can build a software-based model of it (a simulation), you can document it, or both. But for some reason you believe that documentation doesn't require validation, but the tool that expresses a model represented by the documentation does not.
Realistically speaking, they have 2 days between command and response, so they have plenty of time to write a simulator.
Yes, I have strong reason you underestimate the complexity here.
It would have taken much more time than they had available, and since an assembler would be a new tool, it would have required certification. (So, even more time and paperwork.) Plus, they had incomplete docs and there is no working copy or simulator of Voyager here on Earth. So any assembler written would by definition be incomplete or inaccurate.