Hacker News Reader

Comment by Dylan16807 | Hacker News Reader

It's only chesterton's fence if I treat your method as a baseline.

I thought you meant some kind of pass-through licensing, but otherwise the expensive stuff like database licenses generally charge per core, don't they? That's going to scale very directly with your number of customers. If you need 4 web server cores and 1 database core per ten customers, then you should not be treating servers as a fixed cost, you should consider each customer to cost half a core and 1/10th of a database core license fee. It's not perfect but it's much much closer to reality than thinking about servers as a fixed cost.

Don't go buy 150 servers in anticipation of customers you don't have.

View on HN

it's the accounting/finance profession's fence you're quibbling with here, not something i just made up. those are indirect costs. in marginal analysis you only consider direct costs, not indirect ones (for background: https://www.investopedia.com/terms/d/directcost.asp).

"Wages of production staff" are more indirect than server costs if you're getting servers as-needed. Server use is easier to tie back to specific customers than employees, in part because you can add and remove them so easily.

"A direct cost is a price that can be directly tied to the production of specific goods or services." yes that is the case here.

And if I look at the first search results specifically talking about servers:

https://www.ibm.com/downloads/cas/DG7NY5QW says servers are a direct cost.

https://blog.hubspot.com/marketing/direct-costs says servers are a direct cost.

you can't simply classify servers or employees categorically as direct or indirect costs, but must consider how the inputs contribute to outputs. a janitor is a direct cost for a cleaning business but not for most others. if you're aws or ibm, then yes, (fractions of) servers are a direct cost. but servers are too course of an input, and non-linearly related, in most saas businesses (e.g., stripe or hubspot) to be apportioned to individual customers. that's not the same as being able to statically divide server costs by the number of customers to reach a per-customer cost, which you can do with any business and any cost.

take an accounting or pricing class if you're really interested in this stuff. that will explain it much better and more in depth than a few sentences here can.

If you're doing SaaS you can measure exactly how much server is used by each customer. If you choose not to, that's your problem.

And servers do not have to be coarse. They come in all sizes.

In particular I want an explanation of how servers are less direct then the wages of production staff. You can measure output similarly, and you can scale your number of servers much faster than you can scale your number of staff.

i used to tutor mba students on such topics. i'd be happy to address your questions at my standard hourly rate.

Well, you didn't address my logic and you didn't address my sources (the one from ibm is not talking about being ibm), and you didn't link your own sources that mentioned servers. You managed to get downvoted for being so unhelpful.

So even if you're the rightest person in the world, you've done an awful job of commenting. There's no reason to believe you, or to believe that you would actually address what I said if I did pay you. Otherwise I would actually be tempted to do so.

Reply on news.ycombinator.com