Question: could you train a model like this from before a known but important scientific breakthrough happened and see if it is able to work it out.

At least then you know the answer yourself and know it’s something that can be reasonably worked out.

This is an active area of research. Demis Hassabis proposed training a model with a strict knowledge cutoff before 1915, and seeing whether it can independently arrive at general relativity.

I think this is a good way to test a certain kind of capability, but as to whether LLMs would pass such a test, I'm guessing almost certainly not. If you've ever used one for research, it's very 'in' the current literature, whatever that may be. It's an incredible retrieval tool, and it will glibly evaluate any novel ideas that you feed in, but analyses are often incorrect when there's a paucity of directly relevant training data.

This is a really fascinating idea… Just another one for the list of side-projects I’d like to get around to but never will!

A purist just wants it to be about the runner not the shoe.

Purism is extremism about a thing. Pick a thing, be purely about that thing.

I used to love F1 for the tech that would filter down to my car in ten years time, but that is not a thing anymore.

I for one love the advances in technology in something as supposedly simple as a shoe. And maybe I'll get to use it on a hike in a few years.

> I for one love the advances in technology in something as supposedly simple as a shoe. And maybe I'll get to use it on a hike in a few years.

These shoes are practically disposable. They trade longevity for noticeable gains in performance. Even the tier below don’t last very long. This is not tech that is going to filter down to your hiking boots.

There's a whole industry of hiking gear that has been steadily trading weight for durability. I only hike in running shoes (altras) which only last me 400 or so miles. It's way less durable than my old boots.

I doubt this sort of thing will translate, but I wouldn't put it past hikers to opt for something less durable to shave a few grams.

Fair enough, I obviously haven't looked into it.

But the initial tech on an F1 car was not made for 200k miles either.

Even if the full tech stack to make it all work - material science, physical layout and construction - doesn't transfer, maybe some bit of it will.

My point being though, unlike some purists, I like the technology race. It is much better than having a brand war simply on the basis of brand loyalty.

This is not the point

What is?

There’s info in one of the other threads about better carb intake too.

But yeah at this point, “it’s the shoes, stupid” should defo be the main part of the conversation.

These were Sabastian Sawe's splits

5km - 14:14 10km - 28:35 15km - 43:10 20km - 57:21 Half - 60:29 25km - 71:41 30km - 1:26:03 35km - 1:39:57 40km - 1:53:39 Finish - 1:59:30

Yomif Kejelcha also ran sub-two, clocking 1:59:41 on his debut marathon

You have to feel for Kejelcha - breaking 2h marathon and not even winning the race!

Sometimes they have big running machines with a crash mat around them running at 2h marathon pace at running shows. I’ve o ly seen them on video - no one can keep up with it for more than 30 odd seconds. It’s INSANE they are running this fast.

Also bear in mind running a single mile under 4 mins was considered impossible for a long time.

Yesterday I put my AirPod in and squeezed it, expecting Spotify to play - but it must have quit, and instead my mac opened up Music - and that album was STILL there and started to play. How many years has it been?

But it says it’s the same for decaf. That is more interesting

Every now and then I feel like I got a handle on AI and it makes sense. Then something like this happens and either my whole concept of it is wrong, or this makes no sense and is insane.

I wondered this too - why are you being downvoted for asking?

All the difficulty to create that laser it seems fair enough to ask!