Tiny Neural Networks. Bezos Space Habitats. The Cement Ban. The art of noticing. Jellyfish supper. — No.80

Probably of interest to very few readers but just in case; I managed to wrangle up a pass so this week I’ll be attending C2 Montréal, hit me up if you’re around.

This week: Tiny Neural Networks. Bezos Space Habitats. The Cement Ban. The art of noticing. Jellyfish supper.

A year ago: The Key to Everything by Freeman Dyson. (A favorite from last year, now paywalled, search the title in G to get in through that referrer?)

A new way to build tiny neural networks could create powerful AI on your phone

Overview of a research paper proposing that neural networks could be a lot smaller than they are now. Interesting read to better understand these networks and for the whole pruning concept. Not part of the results but it’s a bit disturbing when, in creating these neural networks, the term “lottery ticket” is used and we still don’t know exactly how things work. This smaller networks possibility, paired with the advances in synthetic data, and that some AIs can be based on much smaller data sets for training, points the way (I reckon) to AIs which would require smaller investments and be possible in more places / companies. Promising to somehow get out of the influence of Big Tech but also very disquieting for the same ethical fears and problems we are already dealing with.

[T]he researchers have made a simple but dramatic discovery: we’ve been using neural networks far bigger than we actually need. In some cases they’re 10—even 100—times bigger, so training them costs us orders of magnitude more time and computational power than necessary. […]

Whereas a tiny neural network may be trainable in only one of every five initializations, a larger network may be trainable in four of every five. Again, why this happens had been a mystery, but that’s why researchers typically use very large networks for their deep-learning tasks. They want to increase their chances of achieving a successful model. […]

He believes it would dramatically accelerate and democratize AI research by lowering the cost and speed of training, and by allowing people without giant data servers to do this work directly on small laptops or even mobile phones.

Oh, Jeff

Why Jeff Bezos’s Space Habitats Already Feel Stale

Jeff Amazon wants to go to the Moon but he’s using old ideas, old imagery, and doesn’t bring much new perspectives to the problem (except for his rockets). Fred Scharmen brings the historical context (I mean, he literally wrote the book on this), the O’Neill colonies, the critique, and the irony in this piece on Bezos’ announcement.

His plan to change these trends started with an outpost on the Moon. There, a small number of people could begin a mining operation that would support the next phase—the construction of large-scale, rotating habitats in orbit that would contain reconstructions of Earth’s cities and landscapes, becoming home to millions. […]

Now, in 2019, Jeff Bezos wants his private space company to take over the public imagination about life in space. Bezos is the head of a retail empire, and he knows how to sell an image, but what he’s offering today is a watered-down version of nostalgia for yesterday’s future. Bezos’s proposal is a version of O’Neill’s project that somehow manages to look and feel less futuristic than its predecessor. […]

It’s not just the imagery that’s stale. The framing and assumptions behind the whole enterprise are outdated, too.

Jeff Bezos wants to colonise space, but he’s paying for it by destroying Earth

Amazon is courting and selling to oil companies. Jeff has given up on Earth and is trying to sell us his re-heated space settlement “vision.“

Last year, the company announced plans to run its own delivery services. But unlike FedEx and UPS, which are slowly electrifying their fleets, Amazon ordered 20,000 diesel vans. And, perhaps most damningly of all, the company has aggressively courted clients in the fossil fuel industry, pursuing partnerships with the likes of Shell and BP to provide data services (including one called Amazon Glacier) to oil fields. A page on the AWS website describes how Amazon can help fossil fuel companies “find oil faster,” “recover more oil” and “reduce the cost per barrel”. […]

Enlisting the help of a private company such as Blue Origin may well prove a cheaper way for the US government to achieve this goal, but it means that the benefits – all the science and technology innovations – won’t trickle down to the rest of society as they do with Nasa. […]

Amazon is already sacrificing the climate at the altar of convenience. Blue Origin wants to asset-strip the moon to build refuges for the rich, and it’s paying for them by destroying the Earth.

EARTH: The Cement Ban

Catching up on some podcasts and this one from March by the ever interesting Rose Eveleth is a great listen. Starts from a speculative cement ban, and goes through some of the industrial era history of the material as well as the great difficulties in bringing potential replacements to market, and to actual use. Good example of how much work will need to go into making all the changes we need actually… happen. Loads of links in the notes too.

For the last fifteen years, David has been developing this invention into a material he calls Ferrock. It’s made from a few key ingredients: Steel dust, which is a byproduct from all kinds of industries that generally doesn’t get recycled. Silica, which is basically just ground up glass. And CO2, which is what makes the material harden. And for the last fifteen years, David has been trying to refine his product and get it out onto the market. And that first part, refining his product, that’s actually gone pretty well.

The art of noticing: five ways to experience a city differently

Rob Walker in the Guardian with a short take on his own book coming out this month, The Art of Noticing. It’s straight from the author and includes five things to try for noticing the city but if you have more time, I recommend listening to the interview he gave on Curious Minds (42 mins), loads of great thinking and “practices we can use to become more observant and more connected.” Very much looking forward to reading his book.

Just remember that real engagement with your environment also means noticing things that are not necessarily charming at all: security cameras, potholes, weeds, ruins, irritating strangers hollering into their phones. But sometimes these deserve attention too. Maybe you’ll notice problems that could be fixed, and have ideas about how that might happen. […]

Nobody is making an effort to direct your attention to the ghosts or ruins in any given landscape. But if you want to understand a place more deeply, these are exactly the things you should look for.

Jellyfish supper delivered by drone? Radical future predicted for food

Supermarket chain Sainsbury’s in the UK commissioned a report which includes some input from futurologists and plant scientists, and “explores what we will be eating and how food will be produced in 2025, 2050 and in 2169.” The latter feels quite far off for this kind of exercise (see first quote), the article does include some interesting shorter term forecasts (second).

Sainsbury’s predicts that by 2169 it could be routine for people to hold details of their nutritional and health information in a personal microchip embedded in their skin, which will trigger an alert to the supermarket. It would then deliver by drone suitable food and drink based on their planned activities for the coming days. […]

Future menus will inevitably feature less meat and dairy, it says. In the shorter-term, the report predicts a quarter of Britons will be vegetarian in 2025 (up from one in eight today) and half flexitarian (up from a fifth today) who eat meat occasionally. The “alternative proteins” market is set to swell by as much as 25% with algae milk predicted to be the next plant milk to take over from nut-based versions.

More: The report itself is here.


  • Design and Futures: Special Editor’s Introduction * Journal of Futures Studies. “Part of what bringing design and futures into sustained dialogue does is to allow each field to become more fluent in a second language which is the other’s native tongue.”
  • ? The Training Commission ”is a speculative fiction email newsletter about the compromises and consequences of using technology to reckon with collective trauma. Several years after a period of civil unrest and digital blackouts in the United States, a truth and reconciliation process has led to a major restructuring of the federal government, major tech companies, and the criminal justice system.”
  • ?? Flow State. “Every weekday, we send out two hours of music that’s perfect for working.”
  • San Francisco Bans Facial Recognition Technology “‘I think part of San Francisco being the real and perceived headquarters for all things tech also comes with a responsibility for its local legislators,’ said Mr. Peskin, who represents neighborhoods on the northeast side of the city. ‘We have an outsize responsibility to regulate the excesses of technology precisely because they are headquartered here.’
  • Why the Guardian is changing the language it uses about the environment. “Other terms that have been updated, including the use of ‘wildlife’ rather than ‘biodiversity’, ‘fish populations’ instead of ‘fish stocks’ and ’climate science denier’ rather than ‘climate sceptic’. In September, the BBC accepted it gets coverage of climate change ‘wrong too often’ and told staff: ‘You do not need a ‘denier’ to balance the debate.’”
  • ?? Fascinating. How Does Venice Work?. “The canals, the sewers, the buildings, the bridges and the rest of the Venice’s infrastructure has all been engineered to deal with a particularly challenging environment: not-particularly-solid ground constantly battered by salt water. In this short film, we learn how the city works and what steps have been taken over the centuries to ensure the smooth function of the city.”
  • ? BBC building ‘public service algorithm’. “His hope is that audiences will stumble onto something new, instead of content that simply reinforces their views. Algorithms ‘do not have to create echo chambers,” he added, “they can open them up’.”
  • ⚔️?? Guilty pleasures: quite enjoying those SYFY recaps. Break down Game of Thrones’ “The Bells” with our epic Rap Up.


Some of the articles for this week were found via Doug Belshaw, Jay Owens, and Johannes Klingebiel. Thanks!