Even if you do a lot of work on paper you're going to look stuff up online or in the code base, which is visible on screen.
That doesn't mean that this tool can't be abused in both directions, but it seems like an unfair critique of the GP post.
OP said something about "not seeing anything on the screen for an hour or more". Not multiple hours. I would almost guarantee that Atkinson thought about the Quickdraw problem for more than an hour before coming up with his insight. I know he wasn't "looking it up online" because, you know, that didn't exist in 1982.
If I'm thinking about something really hard while working at home, I go in my bedroom and close my eyes. That looks just like I'm sleeping, but, you know, I'm hard at work. Sometimes I take a shower.
I think the attitude that "typing" = "programming" has much to do with the popularity of languages like Java that encourage the production of reams of code with very low semantic content.
That's a fairly strong assumption and your subthread here would've gone a lot better if you'd started it by asking whether your assumption is correct or not.
He is quite clearly incapable of making an accurate estimate of the amount of time, planning, and preparation involved, or he wouldn't need to spy on the programmer to assure himself that the programmer was "working". He'd know just from the number of billed hours.
In this case, this programmer did good work for a few weeks that we had no issue with. Then his work largely wrapped, and he announced he would continue to monitor our server for errors. He also continued to bill 40 hours a week for several weeks, which surprised me, but at first I thought (as a non-technical guy) perhaps that was necessary.
After a few weeks of him billing 40 hours a week yet not saying anything to us, I began to get suspicious. When I checked his screenshots, I discovered only one or two screenshots per entire week were actually spent on anything related to our server.
As a test, I decided to see how long it would take me to do the same monitoring he was supposedly doing. It took me about 5 minutes to log into the server and run the command the first time I did it, and perhaps 15 seconds to scan the results that came back. We only needed to do this a couple times a week.
It may be he was a totally honest guy who was just extremely, extremely slow. Far slower than a non-technical guy like me, despite his experience (and he was experienced). But it looked a whole lot like a guy who'd completed his project milking the clock to keep getting paid full-time until such time as the employer figured out what he was doing, to me.
The next programmer we brought on simply fixed the error we were monitoring for so we didn't have to monitor it anymore.
I'm obviously a lot savvier about hiring programmers and developers than I was back then (this was one of our first technical hires). The point of sharing the story was to shed some light on what the value to employers is of having this technology. Without the damning screenshots, this guy may have milked the clock a lot longer, and we likely would've been a lot more sour on hiring devs in the future. You may not care if any one individual employer gets burned, but multiply that by tens of thousands who are now less willing to pay good rates for developers, more suspicious of those they do bring on, and keep their devs on shorter leashes, and there's something to be said for employee's side too of monitoring enabling employers to get the bad apples out fast and make sure the good ones have free rein to be fruitful.
How does taking screenshots let you know that the person is thinking?
Okay, yeah, the guy who was looking at "Boob Forest" wasn't working, but just because you don't see a constant stream of keystrokes appearing on-screen doesn't mean a programmer isn't working.
Edit: this famous story comes to mind.
http://www.folklore.org/StoryView.py?story=Negative_2000_Lin...