s2e23: Recreational Capitalism (3); Miscellanea
by danhon
0.0 Sitrep
1.0 Recreational Capitalism (3)
But then, as Aylett pointed out to me, there’s the opportunity to influence what comes after. Which I hadn’t really thought about yet. Or, rather, the extent to which I’d thought about it had been: in what environment do we want recreational capitalism to exist? Do we want recreational capitalism to exist? Do we want as many people as possible to have hobby “jobs” or ways to get “income” out of doing work, but who at the same time are still able to survive on a comfortable living income? Because right now, I’d say that the proportion of “recreational” capitalism in the Uber-and-Airbnb-sense is in the minority, right? The rest of it is going to increasingly become centralized (ie: people who own cars and rent them out to Uber drivers, people who own multiple properties and manage them via Airbnb).
Or, another version of the future where, say, everyone accepts that every resource is going to be efficiently (or better) allocated via a network-mediated market, so landlords let renters sublet via a service like Airbnb in exchange for, say, the now-standard 30% cut of revenue.
Separately, there was a bit in my conversation with Matt Rhodes that stuck out at me as quotable or like a good nugget of something, that: “self-publishing is absolutely an enabler for recreational capitalism”. Rhodes went on to say that “the easier it is to mass-produce something without gatekeepers, the more likely it is that the particular art form has gone over to recreational capitalists.”
To which: this seems to lend itself to things that we can do on our end, first, right? The network made it easier to distribute (clearly), and easier (but still demonstrably hard, due to centralization) to be discovered. But, if you’re doing something on a recreational basis, you have one luxury that the non-recreational capitalists have, which is time. You can afford – because it’s recreational! – to write your book, say, and then just wait and see if it does well, and if that takes five years for it to be discovered (ie, for the right confluence of the right time and place to happen – the chance/lucky bit), then so be it. Or maybe it isn’t discovered. That’s okay, because it’s recreational! So things that can be self-published: RPGs, novels, comic books, plays, art, poetry – they have been made *easier* or more viable to be recreation that also results in the exchange of value. I mean, you can still give them away for free or do them for free. But it turns out that some of them are also good enough for other people to want to pay money for them. So: look for the things that will slowly become easier for individuals to do – because potentially it’s harder to scale *teams* of humans than individual actors – because that’s where more quote disruption endquote is going to happen.
2.0 Miscellanea
I’m playing a bit more with the pet neural networks – I’ve got another one training on 25 megabytes worth of all State of the Union addresses ever (which you can get here[0]), but it’s clear that the 400 neurons worth on my puny little 13in Retina MacBook Pro aren’t enough (and at about 200k iterations at around 1s per iteration, that’s around 2.3 days worth of solid computation) and anyway, I’d want a substantially bigger network, and I still haven’t spent the time setting up a proper Amazon EC2 GPU instance (ie: get Torch and CUDA working properly) to run it somewhere with enough muscle. Which is annoying because between the PS4, PS3 and Xbox 360 in my house, I could probably get it done inside of a day.
[0] https://github.com/danhon/char-rnn/tree/master/data/sotu
—
OK, 11:15pm and we still haven’t finished Pacific Rim yet. Back tomorrow, I hope.
Best,
Dan