(I work in this area of research.) You are right, the paper is about just a sequence-to-sequence transformation model that learns good replies for inputs but is not actually "understanding" what is going on.

At the same time, we *are* making some headways in the "understanding" part as well, just not in this particular paper. Basically, we have ways to convert individual words to many-dimensional numerical vectors whose mathematical relations closely correspond to semantics of the words, and we are now working on building neural networks that build up such vectors even for larger pieces of text and use them for more advanced things. If anyone is interested, look up word2vec, "distributed representations" or "word embeddings" (or "compositional embeddings").

If you already know what word2vec is, take a look at http://emnlp2014.org/tutorials...

Right now they are talking about if the API is actually copyrighted. If it is, there is still a (good?) chance that fair use will allow you to reimplement it anyway; but that's going to be another court case, likely.

TL;DR: US executive shares the appeals court opinion that APIs are copyrightable, but that does not mean the copyright is enforceable - there will be another court case that will be about if it's fair use to re-implement the (copyrighted) API.

Here is maybe the most important paragraph (italics mine):

Despite the inherently functional character of all computer code, the Copyright Act makes clear that such code can be copyrightable. Nothing about the declaring code (API declarations) at issue here materially distinguishes it from other computer code ... . Although petitioner has raised important concerns about the effects that enforcing respondent's copyright could have on software development, those concerns are better addressed through petitioner's fair-use defense, which will be considered on remand.

The brief is quite well readable (modulo the awful scribus ui), try it!

I'm running java-based UIMA pipelines with large NLP models loaded in memory, plus some pandas datasets loaded in ipynotebooks occassionally.

I don't run 128gb because that's tricky to stuff in a laptop and I have better uses for my money at this point than buying overpriced SO-DIMMs too.

I do use the memory, thank you very much. I just use the computer for something else than web browsing too. I do realize it's getting uncommon (and don't actually even get *that* grumpy about it, just have my different set of preferences).

When you see a guy in sibling comment complaining "In addition, I hate the extremely long time for startup and new tab creation, which is accompanied by constant disk grinding." - well, that's exactly the memory problem, which now translates to bad user experience. As you suggested: The OS swapped it out as another app needed it.

Oh and also the fact that middle click inside website does not load URL from clipboard. It works on a favicon - except in case of verified identity SSL servers, there's no favicon.

About the tabs, frankly, for me (but clearly not just me) a more flexible paradigm which blends seamlessly the concept of tabs and bookmarks (and ideally full-text search over my "bookmarked tabs") would be awesome. I'm a pack-rat and would like to archive whole tab trees for later, see them among the other pages, but not take memory+CPU now. I think there's an actual and large market gap here.

No vertical tabs 10 years after widescreen displays started spreading widely?

Also (not so much about UI), if you have many open tabs, chrome eats much less CPU on the background, but is much more memory hungry.

Maybe you have never owned a notebook or a cellphone, so let me note: batteries deteriorate.

> One of the features of safety glass is that when it breaks there aren't (or many) pointy edges created.

Which kind of safety glass?

They were talking about windshields, those are laminated glass. That means you have two sheets of ordinary annealed glass (which DOES break into big, dagger-like sharp pieces) with a plastic sheet in between (which prevents those sharp pieces from going anywhere). Presumably, given an appropriate substrate, you could make laminate out of any glass-like sheet.

The other kind of safety glass is tempered. This causes the glass to be stressed along the edges so that when it does break, it breaks into a million tiny pieces (all of which are very, very sharp). It may also simultaneously pop, especially if hit along the edges. It's less dangerous because the pieces, while sharp, are simply too small to do any real damage even if, say, a piece explodes while you're holding it.

Source: I worked for a cut & temper operation, I've dealt with all kinds of glass.

First, the situation is more complicated outside of math/physics/cs. E.g. in biology, getting papers is much more complicated, which has connections to computer literacy of authors, conventions in the field, and maybe also conditions of the journals.

Second, if you are an institutional researcher, there are good chances that your institution has subscription to the major places relevant for your research.

I hope it's a problem solved over time largely by natural selection. I'm much less likely to cite papers that I can't read easily.

AFAIK Elsevier's pricing structure and conditions on things like arxiv uploads is much stricter than Springer's. Not sure though.

Ok, that makes sense. I use vim only on Linux, so I wouldn't know. Ctrl-v in normal mode is what always does this for me in terminal vim, but I guess it might interfere with clipboard shortcuts on other platforms.

(Then again, the argument sounds as that vim loses because vim has this different across platforms, while, Notepad++ wins because, well, it is not available on some platforms. In Linux desktop environments, alt+drag typically drags the window and is not passed to the application.)

No, my point is that it's often easiest to *understand* things from a couple of examples, especially when your programming fundamentals are solid.

But even what *you* are saying is nothing bad to do, when you just need to do a quick hack - quickly. (The real burden is on deciding when a quick hack will or will not do.)

In the end, it's about whether what you create works. SO helps that happen.

Exactly. When the concepts are already well sorted out in your head but you need to quickly get something going with an unfamiliar API, it's typically way more time-efficient to just peek at a few code snippets over ten seconds rather than plodding through a confusing API docs written by a graphomaniac with ADHD. StackOverflow is a god-send that made me immensely more productive, especially in unfamiliar programming environments.

Really, you start with "Those days are long gone", then talk about "letter-sized printouts at 10 characters per inch"?

To me, 8 is just perfect when viewing on a computer screen, it has the perfect balance of clarity and horizontal space management. But if someone thinks it's too much, that's fine - because I'm using tabs, so they can reconfigure. This is where all the anti-tab arguers shoot themselves in the foot. There are some good arguments against tabs that make the spaces vs. tabs choice non-obvious, but seems like noone on this thread has a clue about these.

(Of course, in Python I just follow PEP and :set et, tw=4, sw=4. A whole-ecosystem convention trumps personal preferences, and Python especially (with its semantic indenting) has good reasons to abolish tabs.)