jns' recent post reminded me of something Alan Kay said. In the US, you can
expect to pay something like $35,000 for a new car. But computers cost less
than $2000. Consumers think of computers as being the same as television
sets. Kay would prefer to spend $35,000 on a computer but doing that would
be obscure. Only people who know what a computer can do want a $35,000 one.
My way around this is to apply the RAID idea to computing. RAID stands for
Redundant Array of Inexpensive Disks. (Drive manufacturers/retailers hiking
their prices have erroneously changed this to Redundant Array of Individual
Disks). A trashy online VM for a year is comparable in price to a
several-orders-of-magnitude-more-powerful/storageful low energy single board
computer at home. To the extent that you want something not physically with
you, tildes fulfill that roll and ex-90s clone corporations x, y and z do
not (but consumers will just do what the sales reps whose jobs they envy
say).
People who have been bitten by Google's asps will say, without meaning a
syllable of it that they are interested in heterogeneous distributed
computing; my understanding is that the speaker is repeating something from
a similarly unlettered native advertising Kubernetes populariser blog. I
lodge this aspersion because of how no one saying that can describe mpi/the
90s to me (which I also don't like).
On the other hand to the normalcy of distributed computing, cluster
computing feels kind of obscure now. If you have been in an academic
setting, you probably know this ironically unironic parallel acceleration
story: Using GNU parallel to start several concurrent copies of a python
script (in time booked on a supercomputer).
tl;dr I would enjoin you to join me in stuffing our available processors to
the gills with private key onion/i2p personal bespoke streaming algorithms
grinding through our universe.
