I worked with a guy that used to say: "I need approval for a $500 hardware buy, but I can call a meeting that would cost thousands of Dollars in salaries any time I want"

I wish companies had a "meetings budget" and had to justify the ROI of each meeting... One can only dream :)

One practice I've found useful, that is more general, was to issue weekly reports breaking down the cost of each activity. That is, I'd solicit a ~5-10 line breakdown from everyone in my team of how they spent their time that week, and I'd apply average costs for their function and tally it up.

So we might see posts like $2k for a planning meeting show up on the report, and suddenly people started paying attention to what time was spent on - both for meetings and otherwise.

And making arguments like the one in your example. It also meant being able to make actual estimates over whether pursuing a certain optimization would be worth it vs. buying more hardware.

Amazingly each place I've done this at, they were initially confused why I wanted to do this, as apparently it was weird and unusual to have engineering issue such breakdowns

Wonder how much "weekly report compilation" would cost :-)

Since you're mentioning OVH, it reminds me of a story I've been told about Octave Klaba, OVH's founder (not sure if it's true or not actually): he is said to have forbidden any meeting "longer than 90 minutes" in his company.

By "longer than 90 minutes", he means that the total time spent by the people participating in the meeting should not be longer that 90 minutes (so 2 participants -> 45 minutes, 3 participants -> 30 minutes, etc.).

This fits your logic perfectly.

That's an interesting way to make the participants actually plan and think about what they're going to discuss ahead of the meeting.

I used to work at Yahoo (a decade ago). At the time every hardware purchase went through a review committee in Sunnyvale. Even for e.g. Europe - I worked out of the European head office in London.

That committee included one of the founders (I can't for the life of me remember if it was Yang or Filo). I had to buy $4k worth of hardware once and they spent 20 minutes grilling me about load data etc. from the current server.

It just baffled me at the time that a multi-billion dollar company saw it as worthwhile for one of its founders to spend 20 minutes discussing a $4k purchase.

Not to mention the signal value about how little they trusted the judgement of their engineering team.

I concluded that it basically was about two things:

1) to keep said person away from more important decision, at the cost of annoying a lot of other people. 2) as an intentional barrier to make people think twice about asking for more. To #2 on one hand I appreciate fiscal responsibility, but at the same time I've seen too often how making people overly stingy often wastes huge amount of resources trying to engineer around something that could be solved for a tenth of the cost by throwing more hardware at it.

"You need to get approval for a $1,000 purchase, but you can scheduled a 1 hour meeting with 20 people without approval"