cernocky

> what the students want to get out of the class, vs what the teacher wants the students to do: There's a huge disconnect in goals and expectations, so there's no way for the teacher to actually win. The fact that there's such disconnect should give the departments pause.

Unless the teacher's role is to scaffold and support the students in acquiring what the students want, gain trust and lower the disconnect.

This is interesting point. I think lot of people assume that training gathers new "metadata" based on the original data and ignore what the training optimises for which is direct copy of the input. Training a model results in a fancy copy paste (unless incentivised differently).

Reminds me of Supernormal font [1] averaged from widely popular fonts.

[1]: https://research.public.services/typography/

one way to think about rock is to acknowledge it as an informational entity. an entity which is likely more passive then lets say a human or an app, yet by simply being part of the environment, it changes what can be done in the environment. if it wasn't there and if it didn't had a certain shape, the opportunities of other actors in the environment would surely be different. after all rock can be used as a tool and even as a computer. if its still not intuitive, think about Aeolian Harp which is a passive statue, yet a musical instrument, or think how you could encode a perceptron or a simple neural net into a stone (through which a water or air would flow for example). now, even if any ordinary rock doesn't exactly encode neural net, it should be more clear that it still affects information flow. does it help?

Ideas are real in the way rocks are if we are concerned with their informational being. They are real informationally - ideas and math participate in forming the world. Nowadays, LLMs, Search and other apps probably affect the world even more than any common rock. Which is more real?

were the students really surprised about the quality of wine in a place called "Swindlestock Tavern" ? :)

I once read a paper about Apache/Meta Thrift [1,2]. Similarly, it allows the definition of data types/interfaces and code generation for many programming languages. It was specifically designed for RPCs and microservices.

[1]: https://thrift.apache.org/

[2]: https://github.com/facebook/fbthrift

What we need is:

1. make science self-publishing using decentralised protocols the default 2. redefine traditional journals and publishers as curators or labellers on top of the network, instead of owners