Not engineers and servers, but time

You know, I dig here on all language sites in connection with work, and very often I come across such a saying: engineers (programmers) are expensive, servers are not. Well, the context of this dictum is classic: they say, the code in our language L.1 is slower by X times than the code in L.2, but it’s more convenient to program (which is often, by the way, doubtful, because of the complex interrelationships within the huge type systems created over several hundred or even thousands of mythical person-years).

So here. IMHO, stupid saying.


')
In this life, time is precious, and the rest is so-so, compensable and renewable. And if, for example, a person spends a week of his time on correctly arranging malloc and free in a program, and gets a result that works 30% faster than a program in L.1, if this program is daily (on L.1 ) it works at least a couple of hours, this week, additionally spent on writing a similar program in the L.2 language for a year of use will save then six months to those who daily wait for the results of this program. And, in general, will leave people more time for other things. And this, by the way, is not an abstract example, in engineering calculations, for example, or among designers, just about such times of code operation. And there is also indexing, log analysis, multimedia coding and the like.

Well, and if you buy more servers ... But you need to take into account that setting up the server also takes time. And server maintenance of additional costs for electricity, cooling, etc. Well, and most importantly, the same time, because the more servers, the more often they fail, the more time it takes to maintain a fleet of equipment in shape. And even cloud servers need to be monitored.

So here. Time should be considered, time, and not engineers and servers one by one. For only time matters. And 30% of the performance loss in the program of daily or long-term (long-term) use is essential.