Re:Problems of Reproducibility in Computer Science
by michaelochurch on Friday December 19, 2014 @04:34PM
(#8774656)
Excellent metaphor. I agree.
No one (to my knowledge) is suggesting "conspiracy" or malicious intent, or even incompetence. Hardware designers know what they're doing, and on the tradeoff between maximizing performance vs. making performance simple, the former is usually the best option.
Re:Problems of Reproducibility in Computer Science
by Umn55 on Monday December 22, 2014 @04:09PM
(#8785242)
It's all about limits, as systems grow in size and complexity and do more computational work you can build more massive systems. But at the same time the programming/maintenance work requires grows.
So you either need
1) a smarter approach 2) focus on whether the problem is tractable or not 3) If you're going to be ambitious you're going to have to pay for it.
Human beings aren't really known for their wisdom however.
It's pretty simple, all code can be modelled as a fluid flowing through a landscape. i.e. imagine a river (which represents code), it courses over the land and there are "cliffs", rocks and hills (representing bottle necks/loads).
The reality is as bus's got wider and transistors were able to be packed closer, it just allow for more water to flow faster. There's no big conspiracy at all. Because software ultimately can be expressed as a circuit.