[BeetleB]

Rates are usually only good when comparing apples with apples. So if you know you'll use the computer for N days, and the alternative is roughly for the same N, then comparing rates and comparing absolute numbers is an identical analysis.

But if you're comparing using a computer for 5 years vs eating cheeseburgers over 10 years, using rates become more challenging, and is prone to faulty analyses.

My point, though, is that cost per use is not usually a helpful metric.

[OJFord]

> My point, though, is that cost per use is not usually a helpful metric.

I think it can be: say you're comparing two cars with the same fuel efficiency, one's more expensive and more luxurious (or to make the same efficiency less contrived, you're considering the 'luxury interior package' upgrade) - roughly how many times (or hours) will you use the car over your ownership; it might be helpful to consider the per use (or hour) cost of that extra comfort.

Or (as a hobbyist/DIYer) a fancier tool vs. Draper/Silverline/Amtech/unnamed clones that'll get it done but won't last nor make you feel warm and fuzzy while using them. How many times will you actually need a ⅑" tartan paint brush with 90deg handle? Maybe it's quite expensive to get the better one that will last for the one time you'll use it. (I'm not very good at that one - know I'd hate myself the second time used it.)

[BeetleB]

> I think it can be: say you're comparing two cars with the same fuel efficiency, one's more expensive and more luxurious (or to make the same efficiency less contrived, you're considering the 'luxury interior package' upgrade) - roughly how many times (or hours) will you use the car over your ownership; it might be helpful to consider the per use (or hour) cost of that extra comfort.

You can get away with this because you are fixing the time interval: You'll presumably drive both cars the same amount, for the same number of hours. Hence a total amount vs per use (or per hour or per year) gives the same result. What the author is doing is comparing per use cost for very different entities, with unclear timelines.

Ask a random person: Is it better to use your phone for 2 cents, or eat an $8 McDonald's meal?[1] Putting aside that both serve very different needs, the question is missing the component: Over what time frame? The phone will die one day, but I'll always have the option to eat McDonald's.

[1] I have no idea where he lives where a McDonald's meal is over $14...

[OJFord]

Yes, I don't think it's helpful for comparing apples and oranges, just for considering whether you want one thing at all, or which among a single class of thing.

An example I've often used before is a Netflix subscription - I wouldn't pay much more for it, but a lot of things look like a bad deal/much lower utility in comparison to it. Pint of beer = half a month's Netflix, for example. It's not helpful, it's a recipe for overpaying for some things and unnecessarily depriving yourself of others.