iiinova — where your digital self keeps evolving

Jeremy Bassis @bassisjeremy.bsky.social · 15h

So it turns out somebody did write a book about part of it and @igsoc.bsky.social is selling it.

120 0

Jeremy Bassis @bassisjeremy.bsky.social · 19h

It turns out that if you talk about the history of ice one too many times you stop getting invited to parties.

120 0

Jeremy Bassis @bassisjeremy.bsky.social · 21h

And it doesn't stop there. The idea that you can make a material less brittle by creating a mixture is widely used in engineering, including in some rocket fuels. So pykrete, as a naval material might be batshit, but pykrete as a recipe to construct strong materials is actually brilliant.

120 0

Jeremy Bassis @bassisjeremy.bsky.social · 22h

There is another part of the story that is important to glaciology. Even if you could solve the manufacturing, refrigeration and insulation part of the problem, they discovered that pykrete, like glacier ice, deforms. You really don't want to build a ship that is going to slowly change shape.

130 0

Jeremy Bassis @bassisjeremy.bsky.social · 23h

100 ft. There is no data in the paper and the number appears to come from a conversation. I can only imagine this is the way the interaction went down. Jeffries (wandering down hall): Yo, how tall are cliffs in Greenland? Roberts: Idk, maybe 100 ft? Jeffries: Cool Roberts: Who is that guy?

120 0

Jeremy Bassis @bassisjeremy.bsky.social · 23h

You can email for a pdf, but there is also a preprint posted of the near final version. We did separate grounded vs floating and even Greenland vs Antarctica.

120 0

Jeremy Bassis @bassisjeremy.bsky.social · 23h

Surprised Allen and @igsoc.bsky.social hasn't jumped in because they are constantly trying to sell a book about Project Habakkuk.

210 0

Jeremy Bassis @bassisjeremy.bsky.social · 23h

Yes! This is exactly the idea behind the "Marine Ice Cliff Instability". If the ice is too thick and starts to break, and the ice is thicker upstream, it could keep breaking. The places this could occur are places like Thwaites Glacier, which gets much deeper as it goes back.

010 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

Ah, but is it? The acceleration due to gravity on Miranda is about 1% of the gravity on Earth (recalling here) so an ice cliff on Miranda could be 100 times taller and experience a similar stress. So actually, 5-10 km is about what we would expect from the same theory!

1811 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

And we still come back to the same issue because Caroline showed that the glaciers with tall cliffs appear to be more "pristine" and lack the extensive crevassing that we see in many glaciers with lesser ice thickness. You can only get tall cliffs if you don't have flaws.

21815 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

Caroline managed to measure the height of something like 15,000 cliffs, which is a remarkable feat. Ice is still a giant heterogenous pain in the ass, but now we have a huge catalogue of cliff heights to explain.

11423 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

This calculation is largely the same that Catherine Walker and I conducted some 65 years later. We, however, had better data. But our calculation still relied on only a handful of glaciers with measured ice thickness.

11102 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

So Jeffries walked down the hall to visit his colleague at Scott Polar, Roberts. He asked how high are the tallest ice cliffs in Greenland and found out that they are about 100 feet. He used this height to estimate the strength of ice assuming they fail if they are taller than this.

31002 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

Disappointingly, Jeffries learned that the strength of ice in the laboratory had huge variability. Again, this is because ice is quite strong if the only defects are related to the crystal size, but can be very weak when you have pre-existing flaws.

11013 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

To calculate this, he needed to know the strength of ice. He asked the physicist Max Perutz, who was involved in the pykrete debacle about the strength of ice. *Yes, this is the same Max Perutz that shared Rosalynd Franklin's data with Watson and Crick about the structure of DNA.

21402 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

The geophysicist Jeffries, however, was interested in a completely different icy question that had nothing to do with glaciers. Jeffries wanted to know how close an icy asteroid could come to Earth or another planet before it disintegrated.

21053 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

The era after WW2 led to a resurgence in glaciological research, much of it supported implicitly or explicitly by the military. Ice was a hot topic with experiments further revealed that the dislocations in ice can lead to creeping flow, partially addressing that earlier controversy.

21072 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

This is because the strength of ice depends on the crystal structure and existence of pre-existing flaws. Alas, the idea of using pykrete was abandoned as unfeasible and boring aircraft carriers of steel remain the norm.

31653 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

It turns out that if you take samples of ice and break them, you get a huge variability in how hard you need to squeeze them before the ice fails.

11172 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

This “let’s make aircraft carriers out of ice” phase generated many of the earliest western attempts to measure the strength of ice in this pykrete project that was called project Habukukk.

31363 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

They investigated towing icebergs from Greenland and then started to seriously question if using an ice and wood pulp mixture called “pykrete”. This idea seems wild, but adding wood pulp to ice does have a well understood if seemingly magical effect of making ice quite a bit stronger.

21586 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

The solid vs fluid question was bitterly fought, but considered a somewhat niche question right around until WW2. With a steel shortage, the British Navy started to seriously investigate using ice as a platform for landing airplanes in their aerial assault on Germany.

11283 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

Some of the earliest questions were about things like, do glaciers move like caterpillars, creeping forward as water fills fractures and expands at night as it freezes.

11221 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

One of the earliest controversies was about whether ice primarily flows like a very viscous fluid or deforms primarily like a brittle solid. Understanding crevasse patterns was central to this question.

11534 0

Jeremy Bassis @bassisjeremy.bsky.social · 16/10/2026

Ho boy, do I have a story for you. The question of how ice breaks has been central to glaciology for well over a century.

6452107 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

My portion of the study was modeling and moral support. Most of the modeling ended up being cut out of the study because the observations ended up being way more interesting.

0301 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

However, for some glaciers, like Thwaites Glacier (aka the Doomsday Glacier), once it starts to retreat it isn’t clear that there are many spots where it can even temporarily stabilize.

2371 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

What does this tell us about the future and marine ice cliff instability? Well, we still haven’t seen any direct evidence of the so-called instability.

1241 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

But--and this is important--the really high calving rates were short lived as the glacier retreated into a region where it stabilized and, in some cases, re-advanced.

1321 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

And what happens when those really tall cliffs start to break? Well, Caroline was able to measure how fast they break and found out that they can break much faster than current parameterizations of calving would predict.

1341 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

What Caroline did was measure cliff heights for an ungodly number of cliffs around the Antarctic and Greenland ice sheets. What she found is that there are only a handful of cliffs that are around 100 m or taller.

1321 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

But these estimates are plagued by assumptions that are hard to verify. And we have never observed unstable collapse so this is something that *might* happen in the future.

1261 0

Jeremy Bassis @bassisjeremy.bsky.social · 15/10/2026

This idea essentially says that once ice cliffs become too tall, they can’t structurally support their own weight and can fail. How high is too high? Catherine Walker, who was my first student, and I estimated the threshold to be about ~100 m back in 2012.

1291 0

Jeremy Bassis @bassisjeremy.bsky.social · 14/10/2026

An example. I have no doubt that given enough observations AI could develop a model that predicted the positions of the planets in the sky. But would AI discover the inverse square law of gravity and explain elliptical orbits? This is the difference between information and knowledge.

100 0

Jeremy Bassis @bassisjeremy.bsky.social · 14/10/2026

I have been thinking about this in the context of my teaching, but this is growing to be a problem in my field. AI is creating a fire hose of information, but this deluge of information is making it harder to generate new knowledge.

130 0

Jeremy Bassis @bassisjeremy.bsky.social · 14/10/2026

The new AI google weather model beats the benchmarks of some of the numerical weather prediction models. AI is currently in an arms race to prove mathematical theorems. Both provide information, but it is unclear if either significantly increase knowledge.

130 0

Jeremy Bassis @bassisjeremy.bsky.social · 14/10/2026

I’ve been thinking a lot about the difference between knowledge and information. AI can produce massive amounts of information. But this isn’t the same as knowledge.

141 0

Jeremy Bassis @bassisjeremy.bsky.social · 11/10/2026

At times, I may have suspected my code was possessed by malevolent spirits but at no point did I consider the emotional needs of my "if statement" nor that my "for loop" was used so heavily it needed unionization.

020 0

Jeremy Bassis @bassisjeremy.bsky.social · 10/10/2026

Look, we were programming neural networks and random forest algorithms back in the 90s and none of us were like "I think my shitty Matlab code has a soul and needs to be treated with more respect"

1142 0

Jeremy Bassis @bassisjeremy.bsky.social · 10/10/2026

Once again, whenever you see a questionable claim about AI, replace the word "AI" with "linear regression". A linear regression can predict the weather? Sure. Seems plausible. A linear regression has a soul? Um, no.

2174 0

Jeremy Bassis @bassisjeremy.bsky.social · 06/10/2026

Celebrate IceCube and, as a polar scientist, I believe IceCube is a phenomenal scientific achievement. But (trigger warning) I also want everyone to read and sit with the USAP report on sexual assault and harassment in the US Antarctic Program because I can't entirely separate those two things.

nsf-gov-resources.nsf.gov

193 0

Jeremy Bassis @bassisjeremy.bsky.social · 05/10/2026

I ran a segment from a paper I am writing through one of those AI detection algorithms and it told me that it was highly unlikely it was generated by AI because of the egregious spelling mistakes, misplaced punctuation and, I am paraphrasing here, unhinged word choice. Guys, I am so proud.

070 0

Jeremy Bassis @bassisjeremy.bsky.social · 05/10/2026

If it is a human, I would take time to gently point out all of the deficiencies and what needs to be done to correct them for future publication. If it is just AI generated slop, I'm not going to spend my time on it.

110 0

Jeremy Bassis @bassisjeremy.bsky.social · 05/10/2026

Most recent reference is from a half dozen years ago, references cited don't support the assertions in the text. No details are provided to explain how the results in figures were produced making it seem like they were manufactured? Figures have legends draw over the plot lines. It is sloppy.

110 0

Jeremy Bassis @bassisjeremy.bsky.social · 05/10/2026

What do y'all do when you are sent a paper to review that sure looks like it was written by AI?

111 0

Jeremy Bassis @bassisjeremy.bsky.social · 19/09/2026

I think climate discourse would be vastly improved if we spent more time talking about the merits of switching from combined stormwater-sewer systems to separate stormwater and sewage systems.

030 0

Jeremy Bassis @bassisjeremy.bsky.social · 10/09/2026

The University of Michigan and Los Alamos seem absolutely shocked that folks don't want a big high performance computing center that does nuclear weapons research and doesn't pay taxes in their town.

mlive.com

Los Alamos, University of Michigan facility draws anger, shouting at Ypsilanti Township event

Concerns about nuclear weapons research, environmental protections boiled over at a town hall on the planned $1.2 billion computing facility.

000 0

Jeremy Bassis @bassisjeremy.bsky.social · 10/09/2026

This seems like another example where people focus on the outcome instead of the process. If I try to solve a problem and fail I learn things along the way and generate new knowledge that applies to other problems.

050 0