Hacker News Reader

You nailed it on the head. "It seems strange to think that companies would leave money on the table, but in older industries I think this is exactly what happens. The money just requires too much organizational churn for their corporate appetite." As companies grow, they get filled with layers of middle management to handle organizational complexity but this is one of the biggest costs centers. Additionally, as time goes by, these are the jobs that are looked at during cuts so everyone in this layer is in self preservation mode. They start hoarding information going up and down to make themselves indispensable and slowly the organization becomes the Titantic and can't get out of their own way.

Off topic:

The more I think about organizations the more I they seem to be the key players preventing change to me. (Which is not necessarily a bad thing)

Business says: We had this great idea, now let's freeze everything in time and let's make money of this snapshot forever, because who knows when we'll find gold again.

Entertainment industry likes to do that especially.

The phrase "exploitation versus exploration" seems appropriate here.

That's not businesses - that's humans. Or do you apply for a new job every two weeks whenever you see a higher salary advertised?

More precisely it's humans when they form a bunch / organization ;).

NO, I'm saying that even individuals are conservative in that they will tend to stay with an existing job, locale or speciality even when an alternative could potentially offer better returns. I don't check the job listings every day.

Risk vs reward. Often the risk is too great, especially for a company that is making a lot of money. Why risk R&D when there's no competitor. Companies are optimised to make money, not progress.

The assumption there is that higher salary is the #1 consideration. My metric is a blend of salary, lifestyle, and risk.

In my opinion autonomy is key. Big organizations like to talk a big game that they empower their employees to take risks. Yet most of the time they default to the status quo. It's culturally there or not.

Exactly. Also they become fat over time shifting the reality of costs.

I think both the 'people' Kay refers to and Kay himself are right. It just depends on the scale you're looking at.

Darwinian processes are near-perfectly optimized ... for an extremely local maximum. Once there, they fail to make jumps to much better nearby maxima.

If you look at a somewhat larger scale, where there are better maxima available, then the correct conclusion is that they are absolutely not optimized.

That's a nice way to put it. My personal opinion is that Darwinian processes are certainly optimization algorithms, they are just not necessarily optimizing for the same thing as what we might think. As an example, survival is a long-term game. I'm sure there were more aggressive adaptation strategies that went for the local minima and succeeded in the short term, only to be surpassed by the more "patient" algorithms.

There's a huge difference between "exploit in the now" and "exploit, but still be viable in a billion+ years".

A great example of this is a virus which is optimized to spread so quickly and aggressively that it extincts itself by wiping out its available hosts.

Wouldn't Darwinian processes produce "just barely good enough" solutions? i.e. once you're at the stage where things are good enough to survive long enough to reproduce, evolution doesn't play much of a role.

A great example of this is me; I'm pretty amazing; I can move around my environment, make a living to a sufficient degree to cover costs of food and shelter, and somehow even have a fiancée and manage to keep another organism alive & fed. However I'm not an expert at everything, I'm pretty bad at most sporting based things (even compared to other, existing humans; not some theoretical concept of perfection), and I spend about a third of my life doing nothing (sleeping), despite this not necessarily being required (e.g. Dolphins can sleep half their brain at a time, allowing them to avoid this inefficient downtime).

Yup. Until there's selection pressure working against you once again. Imagine if society broke down completely, so you had to fight for the limited amount of food that was available?

Somewhat related: it was interesting when my kids were born; they both had to be C-sectioned. I asked the obstetrician about it, and yes, bigger heads are becoming more and more common because mother+child no longer just die when they do occur. We really are removing selection pressures!

The Greedy algorithm at work.

Why didn't Boeing or Lockheed Martin make reusable rockets?

Von Braun originally proposed reusable boosters, although he had in mind a parachute system with a water splashdown. A recovery system for the Saturn V was considered, but it would have added time to the schedule, and the "man/moon/decade" goal would have been at risk.

Alan Kay should remember that evolution includes random mutations that get you out of local extrema, and "interesting" recombinations. The rate of these is maybe low, but there are examples where large organizations renew themselves this way. But I agree with the conclusion that this is slow and unreliable.

There's a paper out there with the title "Cognition is not computation; evolution is not optimization". I haven't read it!

I'm not sure you are accepting the basic principle, which is that evolution is not in any way optimization (not that the evolutionary process is not "optimized").

See something like https://en.wikipedia.org/wiki/Fisherian_runaway

Also Giuseppe Longo's work on randomness as a cause of organizational complexity in evolution: http://montevil.theobio.org/en/system/files/articlepdf/Longo...

The important bit is evolution is not optimizing for things humans understand. If you say evolution is optimizing for more copies of your DNA fragments your closer to the truth than saying 'fitness' in larger contexts. Don't forget there is more biomass in single celled organisms than single celled organisms. In terms of 'fitness' that does not seem like progress.

Optimization implies "teleonomy", the idea that there's a goal toward which evolution is working, however obscure to us. The interesting question IMHO is whether such a goal exists, somehow expressed or implied by the laws of physics. Jeremy England's work on thermodynamic reasons for the emergence of configurations/mechanisms/organisms that are good at dissipating energy is an example of recent research that asks why we have these complex organisms at all, and tries to answer that question with pure physics.

Not really. A fire's heat creates conditions under which fire more easily spreads by drying out the surrounding area. That's does not suggest needs / wants / goals etc it simply a statement of what happens in very specific circumstances. Further, while evolution might seem to be dissipating energy, oxygen + coal is a much larger reserve of energy than you get in similar environment without life.

And the rate of random mutation varies, influenced by evolution. Eukaryotes even have two sets of DNA, mitochondrial and regular, with very different mutation rates. The complexity of biology, I think, is greatly underestimated by most.

If you are more on the side of accepting punctuated equilibrium, then this isn't true. Organisms will evolve to the local optimum and then stay there. By definition, since most mutations will result in lower fitness.

It's only when the environment changes (rise of mammals post dinosaur-extinction) or the organisms themselves change environment (Darwin's finches) that you get the rapid rates of evolution to find the new optimumm.

Some obvious analogies of these two situations in tech are the rise of the Internet (Microsoft did change, but struggled), and the rise of the smartphone (Apple came from nowhere to a position of dominance).

Alan Kay majored in molecular biology and mathematics ... so he probably does understand just how evolution works. (Hint: the Internet is a good source of bios to look at before making assumptions). We put in a fair amount of work many years ago on personal computing and the Internet so that people can be a few typed characters and clicks away from much of the knowledge of the human race ... what does it take to get people to make use of this?

Why didn't we have electric cars before tesla?

Mostly because batteries sucked. Small electronics drove most of the research to improve them.

I guess Tesla vehicles are market leading in several ways, but they aren't a generation ahead of what anyone else can do or anything like that.

We should eliminate the things other agencies can do from discussions like this one. It's not about what other companies can do, it's about what they are doing. Sure, any major car manufacturer could produce a strong competitor to the Tesla, but they aren't.

I dunno, if you've got $40,000 and a tight schedule you probably end up with a Bolt.

But sure, no one is out there doing anything other than Tesla.

And I think I am giving credit where credit is due. Battery research made a lot of progress, Tesla used the resulting batteries to make nice electric vehicles. I mean, they have a battery partner and all that jazz.

Agreed that Tesla likely couldn't exist without battery technology that's developed only in the last several years, however, a Tesla is a lot more than a battery. No company has stepped up to say, "let's build a consumer electric sports car" or "consumer electric SUV" or "consumer electric luxury sedan", and that's probably down to stale corporate culture more than any single other reason.

> It seems strange to think that companies would leave money on the table, but in older industries I think this is exactly what happens.

This is probably true, but it's also deeply heretical for believers in market efficiency.

or evolution...