social.gl-como.it
One of the use cases I thought was reasonable to expect from ChatGPT and Friends (LLMs) was summarising. It turns out I was wrong. What ChatGPT isn’t summarising at all, it only looks like it…R&A IT Strategy & Architecture
There's a trick I pull on my #ADHD brain that works every time.
Whenever there's a task I don't really want to do, I tell myself "okay, we just have to START the task, and then we can stop after <small timespan>" And then, without fail, I end up completing the task, or at least making serious headway on it.
This is because I have a huge problem with task-switching. Once I am ON task, I find it difficult to stop, but I have to get there in the first place.
I hope this never stops working.
I'm not ADHD, but I use a similar trick: I sometimes find it really difficult to do some tasks, but I can often get around them by identifying the smallest step to move that task forward, however little. Like, to pay a bill, my first small step might be to find the envelop that contains the bill. Not open the envelope, not even put it on my desk, just checking that the envelope still exists.
Or find a pen with which I can write "paid" on the bill, once I've paid it.
Auf YouTube findest du die angesagtesten Videos und Tracks. Außerdem kannst du eigene Inhalte hochladen und mit Freunden oder gleich der ganzen Welt teilen.www.youtube.com
A little bit of math can go a long way. The design of a system can be hampered not only by having too little mathematical analysis go into it, but also too *much*.
One familiar example of this is with password requirements in cybersecurity. Mathematically, the more complex that a password is required to be - for instance, by mandating minimum length, special characters, or no reused passwords - the more secure the password becomes. However, make the requirements too complex, and users and service providers will then seek workarounds to the complex requirements, such as easy ways to reset or recover the password, or storage of such passwords in insecure systems - that in fact can serve to *decrease* the security of the overall system, rather than *increase* it. Overoptimizing on just a single metric - the security strength of the direct user/password login system - can serve to compromise the broader objective - a textbook example of "Goodhart's law" in effect en.wikipedia.org/wiki/Goodhart… . Roughly speaking, the security of this direct entry method should be strengthened to be comparable to the security of alternate entry methods, but beyond that any further strengthening tends to be counterproductive. There is little point putting in more locks on the front door of a building beyond the first one, if the windows are unsecured, and in fact doing so may even lead to a dangerous false sense of security. On the other hand, if the windows are harder to access than the front door, then putting at least one lock on the front door makes excellent sense. (1/3)
Even within pure mathematics, we have learned that sometimes the best way to solve a problem is to abstract away information that one would intuitively think to be highly relevant. Much of the progress in analytic number theory, for instance, has been obtained by adopting the perspective that strongly number-theoretic structures, such as the set of primes, should in fact often be treated as much less structured objects, for instance as a rather arbitrary set of numbers obeying some minimal set of combinatorial properties. Abstract too much, and one no longer has enough information to solve the problem; but with just the right amount of abstraction, the problem can move into sharper focus, suggesting the right set of techniques to attack it, and also exploiting the freedom of the setup to perform additional transformations that would not have made sense in the initial setting.
I sometimes like to joke that applied mathematicians need to know the first two chapters of every pure math graduate textbook, but after that the subsequent chapters may have little (or even negative) value to them. On the other hand, the quest to locate Chapters 3-12 are often what made Chapters 1-2 the perfectly polished and useful gems of mathematics that have such broad utility... (3/3)
I really like the thread that you're pulling on here, which aligns with your blog post on "stages of mathematical development", and i think both are true of all specialisms
All progress seems to have two stages, accrue -> consolidate, which repeat (or don't) through time. We tend to confuse accrual (and accrual alone) as *real* progress; when it is the consolidation stage which matters more, because consolidated form is more stable, in that we can build upon it. which is *real* progress
The first stage of specialisation (the second of your post) is accrual, of special details, which we deem existentially significant, and subsequently tend to parse everything by those forms, almost exclusively. But the territory of specialisms is not isolated -- in nature nothing exists alone
The second stage (the third of you post) begins when we learn the plurality of significance of form -- that other specialisms/ ways-to-interpret the same form (or parts thereof) exist; and with each additional interpretation, an essence, the intersect of both, equates to a new generalisation, which is simpler, and which allows us to notice that the surface detail obsessed over by specialists is only part of the whole story
In this way, we can think of all special-domains as delta, to a respective/ relative general-domain, of relative simpler forms
And the interesting thing is that, essences being general, are fewer, and common across different phenomena: so a universal general-domain of essential forms, which all special-domains recontextualise
The implicit point, is that the path to unifying knowledge, is not by union, but by intersect: radical simplification
Which is cool
going to fawn over Standard Ebooks site standardebooks.org for a minute here
So aside from providing beautiful ebooks of public domain books, the site is blazing fast serving pages for so many books, and it's all static pages:
alexcabal.com/posts/standard-e…
Then look on a page for a book. There's a render of a book with the cover, the books at like an angle and if you hover over it the book floats a little.
Turn off javascript and the pic still renders and floats a little. If you right click and say open pic in new tab just the unaltered cover comes up.
Which means all the magic to take the cover and turn into a render of a book with a floating animation is all done in CSS.
And even better, go to a typical book. Typical thickness in that render. Then go to a big book, like The History and Decline of the Roman Empire.
standardebooks.org/ebooks/edwa…
Look at that chonk of a book in the rendering. So each book rendering is rendered the relative thickness of the book. Just such a cool detail.
edit: Oh, and go to the feeds for books:
There's OPDS (the catalog feed format), and RSS and Atom. Normally, you make the mistake of going to a feed URL in your browser instead of a specific program to read feeds, and you're punished with horrible XML.
But click on any feed link here and it's the nice looking HTML like the rest of the site. Is the server determining dynamically if a browser is reading the feed or a different program? But wait, isn't the site static?
That's right, all those feeds are also good looking HTML. Just brilliant work.
AI and the American Smile
medium.com/@socialcreature/ai-…
How AI misrepresents culture through a facial expression.
Turns out that LLM summaries are actually useful.
Not for *summarizing* text -- they're horrible for that. They're weighted statistical models and by their very nature they'll drop the least common or most unusual bits of things. Y'know, the parts of a message that are actually important.
No, where they're great is as a writing check. If an LLM summary of your work is accurate that indicates what you wrote doesn't really have much interesting information in it and maybe you should try harder.
Flow looks amazing :)
made in #blender
#b3d
creativebloq.com/3d/3d-animati…
youtu.be/BdqxF79LPgY?si=-UOaDj…
Flow looks stunningly immersive.Joe Foley (Creative Bloq)
In una precedente news vi avevamo parlato della scoperta, orbita e visibilità della cometa Tsuchinshan-ATLAS (C/2023 A3). Dopo il difficile periodo di visibilità mattutina avuto durante il passaggi…Asteroidi e dintorni
There 👏 is 👏 no 👏 such 👏 thing 👏 as 👏 a 👏 secure 👏 backdoor.
techcrunch.com/2024/10/07/the-…
China reportedly hacked the wiretap systems required by U.S. internet providers under a 1994 U.S. wiretapping law.Zack Whittaker (TechCrunch)
FREAKING FINALLY
"Buttons are back, baby!! This piece has been 15 years in the making, more or less."
@mimsical in the WSJ on the resurgance of physical buttons for interfaces.
"Fundamentally, the problem with touch-based interfaces is that they aren’t touch-based at all, because they need us to look when using them."
Materiality and embodiment ftw.
[gift link, courtesy of the author]
wsj.com/tech/personal-tech/tou…
Allora, che dite, ci vediamo oggi?
#aspassonellastoria #storia #mappe #catastoteresiano #archivioiniziative #gep
It is a truth universally acknowledged that for greater social media engagement, one should share pictures of cats.
bsky.app/profile/zanfa.bsky.so…
“I’m going to buy him a copy of the Mythical Man Month. Actually I’m going to buy him two copies so he can read it twice as fast.”
— Unknown
#Q4TD #Quote #Quotes q4td.blogspot.com/2024/09/im-g…
“I’m going to buy him a copy of the Mythical Man Month. Actually I’m going to buy him two copies so he can read it twice as fast.” —...q4td.blogspot.com
Yesterday was an important anniversary that supporters of #Russia (Soviet Union) do not like to remember:
September 17, 1939 was the day when the Soviets enthusiastically joined the Second World War, on the side of their ally - Nazi Germany.
That's why in Russia they don't like to talk about the Second World War, which they helped start.They prefer to talk about the Patriotic War, that is, the events after they were betrayed and attacked by their former ally.
#USSR #Germany #WW2 #WWII #Poland
Haiku R1/beta5 has been released
Ancient Babylonian map of the world: A 2,900-year-old clay tablet revealed
timesofindia.indiatimes.com/wo…
World News: The Babylonian Map of the World, originating from ancient Iraq around the sixth century B.C., is the oldest known map. Depicting a circular world withTOI World Desk (Times Of India)
"Prima di poter vedere questa ricetta, accetta i cookie nostri e di SETTECENTOSESSANTACINQUE nostri collaboratori."
Uhm no, forse settecento sono un po' troppi, e comunque volevo vedere come fare gli spatzle, non i biscotti.
Grazie lo stesso giallozafferano, ma cortesemente vaffanculo
This article fundamentally changed how I look at weird inauthentic smiles, especially on people not steeped in American culture.
AI and the American Smile medium.com/@socialcreature/ai-…
