#0048: From programming to parenting
I have been writing software or managing people writing software for almost 25 years. Over that time I’ve managed to get a pretty good intuitive sense of what is going on in a piece of software. With a bit of tinkering, some time spent trawling through examples, and some real-world, working code, I can generally bumble my way thru. I can also sit down and draw a picture of the components of a system from a functional or operational perspective, and I can listen to skilled practitioners talk about their systems and more or less understand what is going on.
However, the software world is currently going thru a really profound structural change with the emergence of machine intelligence techniques like deep learning. We’re not there yet, but I can see a time when there will be far less imperative programming in the traditional sense, and far more training, or even parenting of software systems. If this happens in my working career (which I fully expect that it will), then what use will 25 years of intuition, built up around the nuances of imperative programming, be to me?
What are the machine intelligence equivalents of branching, iteration, and invocation? What does debugging look like when you are training a deep learning model? What does automated testing look like? What does version control even mean when you’re talking about a neural network, or something even more sophisticated?
Some of these questions don’t make sense, some are intentionally naive to make the point, and I’ve probably missed out others that are even more relevant. And that’s the point. The future of programming is probably going to be more like training or coaching or perhaps even parenting in the long term, than it is like the programming that I’ve done for the last 25 years.
In many ways, this is a question of intuition. Sam Harris described intuition in a recent podcast as “…when you can break down your knowledge of a thing no further, the step you take is intuitive.”
My job for 2018 is to build up a whole new set of intuitions about the nature of programming.
Speaking of bikes, if this doesn’t win
#coolest-thing-of-the-week award, I don’t know what does: Manta5 Hydrofoil Bike. They say that “submerged starts can be mastered in an hour or so”. Sure. More at manta5.com. I had no idea that these hover-boards were a thing. Check out this powered surfboard version.
If this story can be believed, Anthony Levandowski (who you might remember left Google’s Waymo to help Uber with self-driving cars) has been keeping himself busy by founding a church based on the idea of worshipping a future AI deity. Some parts of his manifesto makes sense, but I’m not sure that it needs to be wrapped up into a religion. Sounds a bit like someone else who once started a religion.
Calls to regulate Facebook (full disclosure: I do not have a Facebook account) seem to be popping up all over the place following shenanigans during the 2016 US Presidential Election. There are genuine questions as to whether Facebook could regulate itself, both from pundits, and from people who have worked inside Facebook in the very area that would be tasked with implementing the regulation. Given that Facebook is built on collecting data about its users, it seems highly unlikely that any internal regulation would work, but it’s also not at all clear that regulators would know how to regulate them. Two possible solutions: 1) stop Facebook from acquiring any more social networking startups, and 2) force them to open access to the social graph (with appropriately fine-grained end-user authorisation) so that new startups have the potential to compete against them.
To wit: Mark Zuckerberg — dead at 32 — denies Facebook has problem with fake news.
There is a pervading narrative that innovation is the purview of the young. Interestingly, when you look at the data, the reality is quite different. For example, even though startups tend to worship the young, research shows people are most innovative when they’re older. In fact, that previous article suggests that the overall median age of innovators is about 47, with less than 6% of the sample under 30. Which is great for me. I still have some time left to come up with a good idea.
I could quite reasonably live the rest of my life without seeing another Hollywood superhero movie. However, this list of the 50 best superhero movies is interesting for two reasons: 1) reminding me that there have been a few good ones over the years, and 2) just how many movies have been made from superhero comics. What on Earth is Hollywood going to do when they’ve finally tapped this well?
Arthur C Clarke’s Rendezvous with Rama, and it’s subsequent trilogy, co-written with Gentry Lee, Rama II, Garden of Rama, and Rama Revealed, are amongst the most widely acclaimed science fiction novels of all time. In this week’s case of
#art-imitating-life it looks like a Rama-esque object of some kind, named “Oumuamua”, has become the first asteroid from outside our Solar System to pay us a visit. Scientists are saying that this is not only super cool, but also totally bizarre.
Only in Australia
Watch this orphaned koala just hugging his teddy bear.