The ongoing multicore parallel computing revolution seems to have gone up a gear or two in the last few years. As I have written before, this is the Age of Amdahl’s Law: what matters is the serial bottleneck.
Work comes in two basic kinds: serial and parallel. The first cannot be sped up by adding new resources but the second can.
In my view, this is the least understood of all economic laws and the one with the greatest consequences for business and social innovation.
Hierarchical decision-making structures are replete with bottlenecks. Think of any bureaucracy. To get a form you stand in line. The first line is to get a number, so you can wait. You wait so you can get in line. The second line is to get the form. However, that form is the wrong form so you go back to stand in line. Then you get a number…
You get the picture. In any society that does this habitually very little gets done.
Of course, this problem is not limited to Government. Business suffers greatly from this problem. So too does Academic Research.
Call me a cynic, but my experience of academic research in physics taught me that before you were allowed to discover something you had to get in line.
The first line was about doing stuff that made you an Academic. Usually that process was so long and drawn out that you wound up forgetting what the line was for.
It turns out you could just go right ahead and discover a whole bunch of things right of the bat. However, then you were Not in Line. Get back in line!
This is a perfect example of a serial bottleneck.
However, once you understand the power of Amdhal’s law then things look different.
Here are the game changers:
1) all physics is an algorithm
2) algorithms can be turned into software
3) software can be protected and is valuable
Finally, in this new world of Google Compute Engine you can spin up a cluster for rent. Hence the capital cost barriers to radical innovation in algorithms is low.
So what are the real barriers?
The social bottleneck of Academic “publish-or-perish” screw-you syndrome pretty much ensures nobody with any good idea will get anywhere.
Chalk that up as a good thing.
The negative network economy of Academic Rivalry pretty much stalls all real advances for at least one generation until all opponents have finally died out.
The Academic flotilla may have all the Pomp and Pretty Pennants of the Spanish Main, but they are: Dead in the Water.
So how does one exploit this?
I think the answer is Privateer Science.
Academics are not crash-hot at business.
For decades, they enriched publishers by madly competing to publish in journals that nobody read. Now they are madly competing to pay top dollar to be published in journals that the public can read for free.
Okay, so first barrier down.
You can read what they did for free while they fight each-other for scraps.
What does a modern Privateer do in this circumstance?
Focus on the profitable bottleneck.
In quantum mechanics, that is many-body theory with strongly correlated fields.
I know, I know, every wanna-be genius in Physics will duke it out on quantum computers.
Okay, cool… but where is the market?
The market for strongly correlated fields is Chemistry, Superconducting Electronics, Photovoltaic Devices and Nano-structures among other things.
These are Big markets, here today, and served by existing Multi-Physics software companies with real product.
And it is all about parallel algorithms
That is what any budding Tesla or Edison of today should focus upon. The Big Score.
Work out a better QED that is simpler for strong self-interactions and you are done.
Better yet, if you find one – keep it a secret. Exit the rat-race of Publish-and-Perish.
Raise the Jolly Roger and load those cannons with grape shot…
All is Ripe for Plunder on the High Seas.