Wait, really Brice Willis and Al Powell?!
Or do you mean Det. John McClane and Sgt. Powell? That makes sense and you’re totally right.
But now suddenly I’m shipping Det. David Andersen and Sgt. Reginald VelJohnson, or is it Lt. Bruce Willis and Capt. Carl Winslow?
Head spinning…
Oh, is this why they say Jesus saves?
It's interesting that the many of the past barons of industry—Carnegie, Rockefeller—competed to be fondly remembered in death by building the biggest legacies and monuments to future American greatness with museums, research centers, and public spaces.
Maybe this generation thinks they're immortal?
They are fantastic transition technology for teens. They can call home, be geolocated, pay for things without cash, listen to their music on headphones, get directions, slowly text friends, and ask Siri questions, but there’s no Web or social media and they turn into regular watches at school.
Is it just me or shouldn't "on three" totally resolve this situation?
And yet it seems at least half the time I say "on three" people go on the beat *after* three.
Am I misinterpreting?
xkcd.com/3232/
And before anyone complains that my calculation ignores all the culture that happens in rural areas: adding rural culture to the calculation will only make the case for State College stronger!
Funny thing, if you ask AI about the cultural capital of the US, it used to answer "State College"! Sometimes it still does if you coax it a little.
All because it ingested this post of mine with a hyperbolic title, which I stand by today!
And also the Pamchenko Twist is such an absurd move it almost breaks suspension of disbelief when they finally perform it. Definitely a wise choice not to add scenes where they might have had to present a wide shot of the whole move and dwell on its legality/impossibility.
Anyway, since we're apparently doing romance/sports crossover films now (we loved Strictly Ballroom, too) we'll do Top Gun next.
Not strictly sports, but Top Gun is a competition so close enough and the film follows genre conventions pretty well. Plus, a true classic and excuse to watch the sequel!
So my guess is this that got dropped in the name of a perfect ending (like the peddler not getting revealed as Genie at the end of Alladin).
[And yes, apparently there are 3 (THREE?) sequels the first of which establishes that they did, in fact, win the gold, but that's not relevant to my point.]
So we know the judges have a good excuse to not give them credit, which is more than they had at Nationals.
So many interesting ways to resolve that! Plus the film is just begging to have a final shot of the medal case (silver? gold? still empty?) in a scene where Kate resolves her daddy issues.
When Pamchenko introduces the move to them he says its legality "is gray area" (great line, perfectly delivered!).
We know they need to nail the Pamchenko and get credit for it to win the gold, and we see the Russian team look on with concern as they skate.
Don't get me wrong: the movie ends at exactly the right place to make it a classic. Leaving the winner of the gold medal unresolved is a good artistic choice!
But the script clearly sets up a tension that goes unresolved:
At Nationals we learn of the big role judges' subjectivity plays.
One thing I noticed this time around (A and I talk about plot structure a lot) is that I'm pretty sure the film ends sooner than it did in the original script.
That's because the movie sets up some pins that never get knocked down, especially with respect to the legality of the Pamchenko Twist.
In our tour of sports films, A (11) and I just watched The Cutting Edge (a.k.a "toe pick").
Pretty corny at times, it's still a perfectly made romcom with great acting and chemistry, AND a perfect Rocky-genre sports flick (his hockey injury is the even same as Rocky's injury that begins Rocky II!).
I think about this a lot with respect to exoplanets, and it all crystalized for me reading Lisa Messari's thesis and book on the topic!
A fascinating insight into the epistemology of science (I mean the actual practice of how scientists decide what is true).
www.dukeupress.edu/placing-oute...
I learned E&M from the 1st printing of the 3rd edition and actually enjoyed it, but yes "romance of style" is not a phrase I associated with the experience!
Anyway, this "somber," "monotone," "incomplete" book that ignores "radio ETI schemes dominated by physics" is available through your academic library for free:
iopscience.iop.org/book/mono/97...
and in physical form here:
www.amazon.com/Search-Extra...
Enjoy!
The reviewer says it has "no romance of style as a DeGrasse-tyson or Sagan" … wow, tough crowd! And also, it's a *textbook*?! 🤷
"Disappointment abounds" in its incompleteness, most notably…2 obscure concepts appearing only in a single conference abstract from 1995 with only 5 citations! 🤣
I know I should never "read the comments" but there's a 3 star review of my textbook that's really funny.
It praises the book for being "suitable for non science majors at a freshman level" which indicates the reviewer has never taught science to undergraduates!
Joanna reminds me of another good take on the subject that came up recently:
Yes, I reserve the right to update my assessment as they get better.
Also, code assistance is a place where LLMs really shine and, in many cases, can improve velocity (of course, so does hiring more programmers!).
But so far the assessments of Claude's leaked source code are pretty awful!
I reserve the right to change my mind as LLMs and generative AI get better.
I'm skeptical they can bootstrap their way there at the moment, though (for a reason similar to the one explaining why pulling oneself up by one's bootstraps is literally impossible!).
I realize some people think that they're priming things for exponential growth but I haven't seen generative AI do anything *better* than experts can. They do some things very quickly and they've done novel things that experts haven't done yet, but is anyone impressed with their code?
Having more interns will definitely get much more work done! And having interns supervise them means that will happen without more labor or time from non-interns.
But also, it's not clear that this is a useful exercise? Also armies of interns aren't free and at some point just hire an expert?
My "treat LLMs like an overconfident and overeager sycophantic intern" mental model continues to seem about right to me.
Learning that Anthropic is proud that Claude is now writing the code for Claude seems to me like being proud that your interns are now each directing a team of interns.
Some problems are hard and only get solved after lots of wrong turns and incremental advances. It's important work that needs to happen and should get documented in papers.
So not every paper needs to be a classic! And not every step needs to be done well or correctly. Science self-corrects.
Another thing that I think bothered me about the Nature article is that it perpetuates the sense that it's a problem that there are a lot of "bad" or incremental papers out there. I tend to be more generous about this, and I explained here:
physicstoday.aip.org/letters/comm...
I hesitate to form very strong opinions about LLMs and AI for now because I know they'll improve rapidly.
In the meantime, I think these two papers are properly framing the problem.
So I agree that everyone arguing that prompt engineering can replace skills is wrong.
That said, prompt engineering does seem very useful to me as a skill (speaking as someone who does not have it) akin to "Google Fu," which has served me very well!