Transmission #32: Dangerous machines, true lies, the perils of organic farming and being bad at guessing.
Design, ideas and other flotsam.
Hello. Welcome.
This is Transmissions by me, Martin Brown. Father. Husband. Design Lead at Craig Walker, sometime lecturer at RMIT. Marty to most.
This is an ongoing fortnightly newsletter that collates some of the more interesting stories, links, quotes and other curios that float my way.
If you’re new here, then sign up now to get more of these in your inbox, and don’t forget to tell your friends!
PROMO:
Get smarter every day
Join 50k+ smart people on Refind and get 7 new links every day that make you smarter, tailored to your interests, curated from 10k+ sources. Link
Design
Dual use of artificial-intelligence-powered drug discovery
Fabio Urbina, Filippa Lentzos, Cédric Invernizzi & Sean Ekins, Nature
Terrifying. A company using machine-learning to power the discovery of new drug compounds, decides to see what happens if they flip the switch and change the parameters from ‘help’ to ‘hurt’. The algorithm then dutifully obliges and produces a string of never before seen and highly dangerous chemical compounds. Is bio-engineering about to repeat the mistakes of the social media platforms, where ‘assuming positive intent’ is just about the most dangerous thing you can do?
Ideas
True Lies
Leo Kim, Real Life
“Conspiracy media gains its power by making viewers feel they control the narrative.” This dive into 21st century media theory is a must-read for anyone seeking to understand the splintering of our social narratives. A key point is the deep sense of truth that forms in people’s mind when they actively ‘discover’ new ideas themselves – “doing your own research” – even when the root of those ideas are, in fact, false.
"Rather than receive a given perspective via the “hypodermic needle” of mass media, in a film like Loose Change, the viewer obtains their conviction through the assurance that they’ve arrived at it independently. Such a belief is harder to displace than one obtained passively."
In Sri Lanka, Organic Farming Went Catastrophically Wrong
Ted Nordhaus, Foreign Policy
An enthusiastic and idealistic Sri Lankan politician boldly commits to a 10-year plan to phase out synthetic pesticides and fertilizers, and incredibly, he follows through. The outcome, only 3 years in, has been a brutal decline in crop yields, pushing the country to the brink of economic disaster, and forcing Sri Lanka to import food to feed its population.
It’s hard to square stories like this with the writing of people like Michael Pollan, who have argued passionately and persuasively against the extractive nature of industrial agriculture.
If the Sri Lankan experiment proves anything, it’s that the planet might not afford to be able to continue with modern agricultural practices, but we might not be able to live without them either.
Chart of the Week
Americans (but not just americans, I believe this goes for pretty much everywhere) are very bad at estimating the distribution of traits in society. YouGov does a deep dive into why, but it boils down to something like: when we don’t know, we guess that 50% is a pretty reasonable figure, and calibrate our guess toward that.
But sadly, even when we’re wrong (and wrong by a lot), having our views corrected doesn’t tend to change our underlying beliefs.
Does correcting misperceptions of group size change peoples’ attitudes on related issues? Current research suggests it does not. In a series of studies (one of which used a survey fielded by YouGov), political scientists John Sides and Jack Citrin attempted to correct inaccurate beliefs about the size of the U.S. foreign-born population, both subtly, by embedding the accurate information in a news story, and explicitly, by providing survey respondents with Census Bureau estimates. They found that while providing this information did somewhat improve people’s knowledge of the number of immigrants in America, they did not make people more supportive of immigration.
Other
👨🏻🏫 Another writer curious as to why we don’t seem to be producing genius at the rate we used to. His suggestion? More tutors. Link
🤖 Deep Learning is hitting a wall, as ever larger data-sets are leading to diminishing returns. Link
💣 Rollicking yarn about an all-women team who hatched the plan to destroy Syria’s chemical weapon supplies. Link
If you liked this, please do share it!