[dreamcompiler]

You bring up a fascinating point. In many large enterprises -- even those that make money in some aspect of computing -- the centralized IT organization controls everything related to computers. I think we would find it odd if, say, the John Deere Corporation was forced by their "Transportation Support" organization to use Honda ATVs for moving people and goods around their factory campus. Or if Yale Law School was not allowed to purchase books written by their faculty. Or if the Stanford EE department was not allowed to equip their buildings with low-voltage LED lighting they had invented.

And yet we just accept that a leading computer science department at a major university can be forced to use crap enterprise software that's vastly inferior to anything they themselves could have written.

Some of my examples might be hyperbolic. But the power of IT departments to mandate a dumbed-down status quo still seems very weird to me. I believe it's one of the factors that keeps computer science and engineering from making more forward progress.

[bberenberg]

I think your examples actually show exactly why IT controls everything, and not people who want to do stuff because they feel like it that day.

John Deere does not make ATVs, they don't make any tier of people moving equipment. The idea that they should use their own is ridiculous. If they didn't enforce the Honda rule, people would be riding around in the front of dozers.

Much in the same way that if Stanford invented new bulbs, they would be used in a lab. With safety standards applied. Why don't you want these newly invented bulbs used all over your building? Well, what happens when they burn your building down? What happens if the people in the lab who invented them decide to make a company selling them, and are busy with that, and now you need to pay your maintenance people to deal with these new bulbs they don't know how to use.

Being slow to change in a larger organization is a feature, not a bug.

[dreamcompiler]

> John Deere does not make ATVs, they don't make any tier of people moving equipment.

They most assuredly do, which is why I came up with that example.

https://www.deere.com/en/gator-utility-vehicles/

[bberenberg]

You're right, I didn't see these when I checked their site.

At the same time, I stand by my argument that there is often a great reason why you want to standardize. For example, do you want a cyclical dependency in your production stream. If there is a defect in your people movers, and you need those people movers to operate, you now have to split the newly produced people mover parts for fixing your production equipment vs getting them out to customers.

The point I am making isn't that IT is some bastion of brilliance and operational excellence. They're mediocre at it. And this is a good thing, not a bad thing.

As orgs scale, you want to be less nimble because any given success or failure is amplified. If a 10 person company screws up and goes out of business it sucks but it's not a big deal. 800 people? That's enough to get a presidential candidate to visit your campus to speak about the important of retaining jobs.

People underestimate the impact of the work we do in tech. Another thread on HN today pointed me to https://medium.com/better-marketing/pepsis-40-billion-typo-c... which I think is a great example. A simple software bug led to 18 million in loses, huge brand damage, and deaths of people involved in the protests.

[dreamcompiler]

I understand your point. There's value in standardization. But there's also huge value in eating your own dog food. When IT prevents dogfooding entirely, it's gone too far.

[bberenberg]

I think we agree. The hard part here is to figure out where to strike that balance.

[lostlogin]

‘The tail wagging the dog’ is what I think when I read your comment.

Thanks for writing this.