HN Explorer

How do you convince management to spend on fast computers?

5 points by greggman ↗ HN
I have friends who works on a billion dollar video game franchise (as in it's a popular game). They work with terabytes of assets and a large code base.

But, their CTO gets them relatively crappy computers. 4 cores, 8gig, 250gig HD (not SSD). He even wanted to get them 20inch monitors but was able to be convinced to get 24inch.

Every time they build minutes of productivity are lost to waiting. The longer they wait the more likely they are to get distracted browsing the internet.

Assume they get paid $120k a year (low). $120k a year is $1 a minute. Assume they'll use their machine during 1 project (~2years) That means from one POV for every $1k spent on a machine it only has to save 10 minutes a week or 2 minutes a day to save $1000.

How do you convince management to spend on machines?

5 comments

[ 3.0 ms ] story [ 19.8 ms ] thread
brudgers
You have to convince them that machinery is the critical path - and in a large organization it probably isn't. The bottle neck is probably elsewhere.

Suppose team leaders meet once a day to review progress and assign workloads - that cancels out at least part of any two minute time gain (and perhaps all of it).

And the larger the organization, the less effect speeding up development may have - adding version 2.5 features to version 2.4 is only beneficial if marketing and product packaging can coordinate their collateral with the new scope AND there are additional new features in the pipeline which can support the release of version 2.5 - and doing that requires approval at increasingly higher levels of management.

Or to put it another way, faster machines may simply mean that the programmers have more time to surf the internet while requiring additional capital outlay. Not that that's the entire story - faster hardware is often an easy way to confer status more cheaply than a raise.

greggman
How does wasting 60 minutes in a meeting cancel out 2 minutes saved elsewhere? I'm not sure I follow that logic. You still got 2 more minutes of labor from that programmer than you would have otherwise.
brudgers
Where did sixty minutes come from?

What happens in organizations is that progress is made in quantum steps, not continuously. As you go higher up in the organization, the quantum is a larger period of time and two minutes of programmer efficiency per day never moves the shipping date forward or changes the feature set.

The only real tipping point from a management perspective is if the improved efficiency eliminates a full time employee.

greggman
Sorry, 60 minutes was my personal thinking of "programmers attend a meeting".

I'm still not sure I see your point. You can pay programmers $1-$2 a minute to do nothing or you can pay them to do something you need done. If you get them a slow machine such that 10-30 minutes a day they are sitting around waiting for a compile, link, or other long build step you're paying them to sit around. Seems like a waste of money. If you can reduce that 10-30 minutes day just a few minutes (say 8 to 24 mintes a day) at least some of those minutes you're already paying for are actually going toward what you want them to go toward (work)

debacle
You've already figured it out - math and financials.

Do the math, show statistics. If they don't care about the downtime, it happens.

It's also worth noting that, depending on the age of the CTO, "4 cores, 8 gigs" might seem like enough for anybody. It seems like a hell of a lot to me - I still can't believe that we have mobile phones four times as fast as my first PC and ten times as fast as my first computer, with more RAM than I ever had harddrive space. I'm not saying it's not needed, but man we've come a long way.