A theory is not useful if not predictive: great new waves of computation are built on math and engineering, with a blind spot for the bikes part.
Find a great new tech platform that’s messy and cutting edge? The math and engineering comes first. The impact comes with the bikes.
Conversation
Let’s take the web: math and engineering built it out in the 90s. But it wasn’t until we discovered that what humanity wanted to do with it was not share academic papers but to ~gossip (social media).
Exercise for the reader: where is ML today?
3
3
Replying to
Your engineering layer is one abstraction level too high. Hardware is the primary engineering driver. The software later is a mathematical abstraction/fiction on top of hardware and can be combined with the math part.
1
2
The biggest advances come from realizing the hardware envelope has expanded in an interesting way, and learning to exploit it by wasting/conserving hardware resources differently.
I mean, you basically innovate against Moore’s law, programming is a derivative layer
1
1
Replying to
This is a real thing, but there is still a distinction between C and Lisp. What would you call them?
1
Replying to
Mathematical aesthetics perhaps? Language design is underdetermined by all 3 elements (math, engineering, bicycle), that’s why there’s so much diversity there.
Bicycle is a good observation. In future it might just be applied neuroscience/biology.
2
2
Replying to
Language design doesnt feel underdetermined to me. You are right that there’s further variety, eg syntax.
these lineages all interbreed. Eg lisp *is* a bicycle for the programmer that uses it.
But some breeds are clearly of a kind. Chihuahua vs Great Dane.
1
Replying to
Individual languages may not be within their paradigm, but across languages with overlapping capability? That’s clearly undeterdetermined. You can pick a 1000 ways to output ‘hello world’ and there’s no “right answer”
1
1
Replying to
Dogs are underdetermined by “hunting, house pet, defender”. Doesn’t make the classifications useless.
1
Replying to
My point is it is a secondary “interior” phenomenology, not a forcing functions defining the bounding box of constraints. Ie, there are no fundamental constraint laws there. In engineering you have quantum limit, in math you have P=NP, in bicycle/ux you have 2 eyes/10 fingers
1
2
At best you have constitutive laws, like brook’s law or the idea that strongly typed languages are safer etc.
Replying to
Maybe. Some bicycle ideas came before the hardware (via physics constraints) would allow them at scale. Some of Apple’s (and others’) ideas couldn’t make it into reality until the hardware could make smartphones.
1

