In my case, it was just code running on a single cloud instance. For someone a bit upstream from me, it could have been a commercially made device doing its stu

In my case, it was just code running on a single cloud instance. For someone a bit upstream from me, it could have been a commercially made device doing its stuff, with the details being trade secrets of its manufacturer. Computers have made science boring to watch. Many things are now automated and run 24/7. Even if you happen to be on site, there may not be much to see. In many situations, automation also makes doing things on your own easier than trying to determine if someone has already done similar experiments. If you are doing exploratory work, there is often no established terminology for the topic. Even if a report of prior experiments exists, it can be difficult to determine if it's discussing the same thing. (AI could plausibly help here.)