Comment by roughly

1 day ago

The thing that is missed in most efforts to replace people with machines is how often the people that are being replaced are on the fly fixing the system the machine is intended to crystallize and automate.

This is what a lot of people miss about "AI will replace" programmers narrative.

When converting from a traditional process to an electronic one, half my job is twisting people's arms and playing mind reader trying to determine what they ACTUALLY do day-to-day instead of the hypothetical offical, documented, process.

Some of the workarounds that people do instead of updating the process are damn right unhinged.

  • Without going into details, just recently I was able to get pretty decent business requirements from group manager, but it seems the only reason I was able to get somewhat decent idea of what they actually do, is because there was certain level of trust since we worked together previously so there was no need to bs one another. I openly stated what I thought is doable and he seemed to understand that I need to know actual use cases.

    edit: Otoh, my boss is kinda giving up on automating another group's process, because he seems to be getting a lot of 'it depends' answers.

    • I will say, in a lot of cases, they aren't BS-ing/lying with intent. Just the general way their minds work seemingly isn't compatible with the very idea of laying out the process in its entirety (inc. the warts/hacks/workarounds).

      So what ultimately winds up happening is, you'll roll out the process according to the official way, and then it is drip-drip-drip of changes as you find out the real-world version.

> how often the people that are being replaced are on the fly fixing the system the machine is intended to crystallize and automate.

If the system is broken, this is actually a good thing.

I have some experience doing automation work in small and large scale factories. When automating manufacturing work you almost always discover some flaws in the product or process that humans have been covering up as part of their job. These problems surface during the automation phase and get prioritized for fixes.

You might think you could accomplish the same thing by directly asking the people doing the work what could be improved, but in my experience they either don’t notice it any more because it’s part of their job or, in extreme cases, they like that the inefficiency exists because they think it provides extra job security.

  • > If the system is broken, this is actually a good thing.

    And the system is always broken. Reality is messy, systems are rigid, there always has to be a permissive layer somewhere in the interface.

    • So many websites and apps are still broken in so many little ways. Maybe broken isn't the right word. But all kinds of annoyances and breaches still happen all the time.

      I generally don't complain/review, and just learn the workarounds/shortcuts, but I very much welcome the increased (albeit perhaps less skilled) workforce leverage, because I think in a year or so we'll see steady improvements accumulating.

  • > If the system is broken, this is actually a good thing.

    Sometimes when you reveal extensive noncompliance with dumb requirements, the requirements get less dumb. Other times, the organisation doubles down and starts punishing the noncompliance.

    My employer's official security policies say everyone should kensington lock their laptop to their desk at all times, even though the office is behind two guards and three security doors. Nobody does. But if someone made a load of noise about it, there's no guarantee they'd remove the widely ignored rule; they might instead start enforcing it.

    • And so people learn to not make noise. And another broken system remains entrenched, forever.

This is exactly why “automation” hasn’t taken _that_ many jobs. It is a totally overlooked detail. Thanks for the reminder.

  • Some industrial shipping docks can be managed by a very small crew. I think this is the metaphor for what's going to happen to a lot of industries.

    • I’m not so sure. They operate that way because of scale and economy (and tech that enables that). In a future where all industries are optimized in such way, very little will actually flow as most won’t have the money to buy goods, thus factories won’t make goods, thus shippers won’t ship, and the global economy grinds to a halt.

      We need waste as much as we need investment. The trick is to find the value in between. I think the sweet spot will be augmenting work, not necessarily optimizing it.

      3 replies →