> He has decided he hates Wall Street and wants to work at a tech startup (good!).

> How can smart, ambitious people stay working in an area where they have no long term ambitions? I think a good analogy for the mistake they are making can be found in computer science.

It's as if being an entrepreneur is the end-all, be-all of all careers. It's hard work. It's riddled with failure. It's not for everybody.

In addition, the algorithm that the author attempts to describe is called Simulated Annealing: https://en.wikipedia.org/wiki/Simulated_annealing

For the system, it's like the particle swarm algorithm (PSO). Well, fine, the system will do fine. But it's really hard on ants.

EDIT: Almost all multi-agent exploratory algorithms have these characteristics: evolution, free-market capitalism.

And we all (as the species, as the beneficiaries of the system) benefit, even if 10M sperm died to give us 1K eggs to give us 1 living turtle to get back to the beach to re-spawn. The species will live. The ecosystem will adapt and thrive -- but it is very wasteful in terms of point-probes on the multi-dimensional problem space. 1K:1 or 1M:1 die, to have the one survive, because it found the niche in the problem space. That's all fine and dandy until we realize that each of us are the point-probes. We think that we are better than that. Well, maybe we are. In 1000 years, they will look back and see.

EDIT2: > It's as if being an entrepreneur is the end-all, be-all of all careers. It's hard work. It's riddled with failure. It's not for everybody.

Well, for the system, it's great. The system searches for it's maximum potential. For the entrepreneur, he's just a point-sample (in one algorithm), or a geodesic field in yours. These things are tried and tossed away as if they are trash if they don't "fit".

Nothing wrong with that ... it's as if the universe is using the human geoplex as a hive-mind to achieve something greater. But, we, as individuals, might have opinions about how we want to be treated as agents of this algorithm.

Looking up PSO, I came across Russell Eberhart's name as the co-author of the algorithm. He is also the author of the excellent book called Swarm Intelligence where I first encountered the Simulated Annealing algorithm (and many other cool ones)

I first came across it when I was hitting certain boundary conditions w/ GA/GP in the early 90s. It's pretty great. Of course, as stochastic methods go, you never know when they are going to get there. I am gratified to find recent research in the DL community going back and saying, "you know what? some of these GA/GP/PSO critiques of multi-layer NN were right, and they had something to say!". But, I'm too old and grumpy to do anything with either gratitude, or revenge or research dollars now. I just mostly tell the kids to get off my lawn.

EDIT: I'm simply saying that saying: "I told you so!" 25+ years later doesn't get me my life or my wife back. I was a misplaced, from my agent-centric-point-of-view, sample. If I could only learn to lean back and give it up for England. (just kidding)

> In addition, the algorithm that the author attempts to describe is called Simulated Annealing: https://en.wikipedia.org/wiki/Simulated_annealing

And the last one i thought was a nod to genetic programming.: https://en.wikipedia.org/wiki/Genetic_programming

You mean Genetic Programming or Genetic Algorithms? Because they're not the same. The former is what happens when you apply the latter to optimize an abstract syntax tree :).

What makes a "true" entrepreneur, as opposed to a "fraudulent" entrepreneur? I think you'd be disappointed to learn why most people start companies.