3000 node clusters? That is a tiny cluster as clusters go. 7 quad core machines? Two ThinkPads? A million dollars on
one-of-a-kind hardware?
You my friend, have zero clue in this. Since you have no idea what the specific cost benefit analysis on this project is that statement is built on nothing but your prejudices. I grant that it may "sound" ineffective to someone with no experience working on such projects. If I were to blather about whether a drug development proposal (something about which I know next to nothing) is "cost effective" without even looking at the financial documentation, I'd come close to what you are doing with that statement.
"I'd be more productive with a personal helicopter with landing pads in various locations with an oncall pilot"
In the right situation, expensing a personal helicopter would be trivial. What is "cost effective" depends on the derived benefit. Applies to startups too. The cost depends on the benefit derived not on what the raw materials cost.
So (fwiw, I could care less about convincing you) It is not really "cost ineffective". Machine/networking etc costs on this project are trivial compared to the total budget (spent on other resources) and (much) more importantly the derived benefit.
(just making up an example since I can't talk about what this project really does) If you were able to decrypt, and classify (in the Machine Learning sense) all enemy communication in real time, how much is that capability worth to a nation? Worth spending a few millions on? Not every project in the world has the financial structure of two guys in a garage coding up the next silly web app.
Iow you are talking about "unknown unknowns" with nothing to back up your conclusions. Amusing but not really relevant.
I see you are making up scenarios where people randomly order new hardware and that too every two weeks just for the hell of it. My assumption is that we are all professionals who won't do crazy things like that on our employer's dime and are talking of what hardware we require to do our jobs well. nostrademons's reply above provides a good justification for a laptop, for example.
No I can't justify helicopters (which was something you injected into the conversation). But I can justify some bandwidth on a satellite, say, and a few other things. Whatever it takes to do my job well. And I can also imagine some jobs ( a general in charge of an invasion say), where the cost of a helicopter on call would be so small in the overall context as to be non noticeable. The point is "cost effective" is always in reference to the overall context.
"you clearly don't work in the same space as virtually anyone here, much less a company like Microsoft."
I never said I did. And in my gp post I was responding to a sneering question as to whether (specifically) I could justify an extra machine or two where (specifically) I work.
I quote "So what you're saying is that where you work everybody has 2 desktops, 2 laptops and nobody would raise a brow if you wanted another?"
Guess what? That's right. Nobody would twitch an eyebrow let alone raise it.
Now wrt Microsoft specifically, even if MS were to give every single developer an extra laptop it cost much less than one of their VPs takes home as an annual bonus for doing nothing in particular. If Google can give their devs a MacBook Pro MS certainly can. (well, not a MacBook maybe but a top of the range Windown laptop). Without blinking.
"how much is that capability worth to a nation?
A fair bit of money. Although you clearly don't do that, otherwise you wouldn't even suggest it on HN. :-)"
Yeah that was a made up example. I explicitly identified it as such. I was just trying to demonstrate that what is "cost effective" depends on what you are trying to do. 1 $ maybe too much money. Or a million dollars may be a drop in the bucket. It all depends.
I do work on similar(ly ambitious) projects. Hardware is (so) not the bottleneck we have. " everyone on this project gets all the hardware they want" just happens to be the truth. Just append " that is needed to do their jobs well". I thought that wouldn't need to be explicitly stated here on HN, but whatever.
(And I am done with this thread. Too deep. should have triggered my cutoff alarm). Apologies to the rest of the folk on HN. Mea Culpa for getting irritated with a (perceived) sneer.
I am disgusted with myself for responding to obvious baiting and will leave the thread here (vs deleting it) as a monument to my folly.
What specifically sounds "cost ineffective"?
3000 node clusters? That is a tiny cluster as clusters go. 7 quad core machines? Two ThinkPads? A million dollars on one-of-a-kind hardware?
You my friend, have zero clue in this. Since you have no idea what the specific cost benefit analysis on this project is that statement is built on nothing but your prejudices. I grant that it may "sound" ineffective to someone with no experience working on such projects. If I were to blather about whether a drug development proposal (something about which I know next to nothing) is "cost effective" without even looking at the financial documentation, I'd come close to what you are doing with that statement.
"I'd be more productive with a personal helicopter with landing pads in various locations with an oncall pilot"
In the right situation, expensing a personal helicopter would be trivial. What is "cost effective" depends on the derived benefit. Applies to startups too. The cost depends on the benefit derived not on what the raw materials cost.
So (fwiw, I could care less about convincing you) It is not really "cost ineffective". Machine/networking etc costs on this project are trivial compared to the total budget (spent on other resources) and (much) more importantly the derived benefit.
(just making up an example since I can't talk about what this project really does) If you were able to decrypt, and classify (in the Machine Learning sense) all enemy communication in real time, how much is that capability worth to a nation? Worth spending a few millions on? Not every project in the world has the financial structure of two guys in a garage coding up the next silly web app.
Iow you are talking about "unknown unknowns" with nothing to back up your conclusions. Amusing but not really relevant.