Thinking On Paper

Mining asteroids sounds crazy.

Mining two-kilometre-deep deposits on Earth forever may be crazier.

Mining asteroids for platinum sounds kind of insane until you listen to Matthew Gialich. Then it feels like the most logical response to a resource problem humanity could use right now.

Astroforge is building spacecraft to mine metallic asteroids for platinum group metals.

And they mean business.

Thinking On Paper got over 1,500 reactions on posts last week! Thanks to evey single one of you disruptors and curious minds. With your support we can all move towards the most interesting technology podcast this side of Pluto. Which remains a planet to us.

Rory Daniels from techUK argues that entrepreneurs can underestimate what the UK already gives them.

The UK has a strong talent base, deep research infrastructure, one of Europe’s leading startup ecosystems and a dense network of technology clusters. It is small enough for founders, universities, investors and government-backed institutions to stay connected, while sitting next to the much larger European market.

The case for staying is not sentiment. It is access: talent, R&D, capital, customers, institutions and proximity.

Born in the UK and operating globally, Wayve brings together advanced sensors, LLMs, connectivity and robotics into one autonomous vehicle system.

“Pioneering a new approach to assisted and automated driving, building end-to-end Embodied AI that scales across any vehicle, anywhere.”

And you might have seen them in London. Its vehicles have been testing there for years.

As Rory reminds us, the UK already has many of the ingredients for a healthy tech sector: universities, research institutions, government-backed science and technical depth across AI, robotics and connectivity.

The challenge is joining those pieces together, applying the right business models and scaling companies that can compete globally.

Can the UK do it?

Let’s hear your side of the argument.

Error mitigation corrects statistically after the computation. Fault tolerance corrects errors while the calculation is still running.

That distinction changes the scale of the machine. If errors are only handled after the fact, they can accumulate through the circuit. If they are corrected as the computation proceeds, the system can run much larger circuits.

Oliver Dial says that is the basis for the 20,000x jump IBM is targeting with Starling in 2029.

Junk food is engineered for the taste buds, not the body. It exploits cravings that made sense when sugar and fat were scarce, then becomes harmful when those signals are saturated.

Carissa Véliz thinks that technology can work the same way. It can be designed around what we crave rather than what serves us, exploiting old human instincts in an environment where stimulation is constant.

Do you agree?

AI philosopher Carissa Véliz uses the Thanksgiving turkey to explain a basic error in how people think about risk and AI.

The turkey is fed every day. Each day appears to confirm the same lesson: the caretaker is benevolent, the system works, the future will resemble the past.

Then Thanksgiving arrives.

Humans can do what the turkey cannot. We can reason beyond the available data. We can ask who benefits, what incentives are operating, and how a system that feels safe today could turn against us tomorrow.

And in there is a lesson about technology. Do you see it?

The US is lithium hungry and supply starving. It has one operating lithium mine.

Jennifer Dunn's research group has been studying what could happen next. In one analysis, they looked at 117 planned lithium mines in the US and what that expansion could mean for water stress, especially as the climate changes.

The clean energy transition needs lithium. But the geography of that supply doesn't always agree. Where the mines are built, how much water they require, and what happens in regions already under pressure? Yep.