Vertical farming is outrageously expensive. If the world has to depend on them to survive, we are doomed. The sweet spot between open-field farming and vertical farming is greenhouses. A lot more control than in the open, but much, much cheaper than going vertical. Israel's farming experience is very valuable, both when it comes to greenhouses, drip irrigation (in the open), collection of dew from the air for irrigation, etc. To feed the world we need not vertical farming but open-air or greenhouse farming, with clever (including robot-based) water, nutrient, weed, and harvesting management.

Octave is not good enough at what Matlab does and not good for anything else. For any new data project I would use Python (with numpy, scipy, and matplotlib). It is very good at visualizing things, has lots of functionality, similar enough syntax, starts fast, and can do a lot of things beyond numerical computing which can become important when the program scales up.

LIDAR does make the system more reliable in good conditions. It is true that it degrades in bad conditions. I would guess proportional to how bad conditions are. Cameras would degrade as well as they also use visible light. Cars also have RADAR, which should see through rain, snow, fog, and smoke, but at a reduced resolution. I guess a car in such circumstances should slow down till it thinks it is safe enough. Car-to-car communication (very easy to implement as an add-on to any car) should also allow cars to negotiate a driving strategy even in poor conditions.

True, Python also uses Perl's regex. But you see, _all_ of Perl reads as a regex! Python at least cleaned up everything else except pattern matching. Oh, the $, #, %, ~ and other such ASCII noise. I miss them sometimes.

I agree Python and Perl are equivalent, and yes, Perl still hangs on. But that's about it. Python in my usage superseded not just Perl, but also Matlab. The whitespace indentation is the only thing I still find awkward.

Perl was my love and I still write code faster in it than in any other language. Yet, the language has always suffered from the fact that the code is hard to read and hard to maintain. Back in the 90's one could put up with that as it was a huge improvement over C, awk, and shell, but no longer.

Perl 7 will carry on the same problems, with the additional complication that there is no backward compatibility. The Python 2 to 3 switch showed just how painful that can be. Keeping compatibility is very important. Spending weeks upgrading software because some clever folks wanted to break things in the name of beauty is not fun.

Now we have Python, which is very mature and can do anything Perl can do and is much easier to mainain and with a lot more packages by now. And Go and Rust catching up, not to talk about a lot of other languages.

You are pushing things here to a logical extreme. To understand the world reasonably well a machine would need to know a lot of concepts, their relations, and some meaning, with physics and other simulations being used for some of that. Philosophy can wait.

Yes, that is a well-known model. The whole point is that it has limitations. If you tell it "a glass jar fell from my hands but luckily hit the carpet" it won't know how to finish that sentence.

Cyc is a valuable project. Yet it is still about how words are logically connected to each other, as I see. That can be enough for some things, but not for all. Adding in physics brings a new dimension, and some statements about the physical world can be understood only after figuring out the interplay of the objects in that statement according to the laws of physics. The next step is, of course, embodied cognition, but even a basic simulator can help.

And of course Cyc is hand-crafted. Translating statements into a physics-based system and drawing inferences from that could be automated and would capture finer-grained detail than what Cyc can give.

Recently there was some news about Google's latest monster natural language processing neural network trained on a multigigabyte text corpus. It was pointed out that the associations one can learn this way are shallow and eventually things can drift into nonsense.

It would help, as in this article, to have at least a basic physics simulator as part of such an AI program, which would give meaning to words like "a ball is on the floor" and "it is raining outside".

Each time the user says something to the machine it would gradually flesh out a physical (or conceptual) model of what the user is saying that lives in some space. This would add coherency and deepen the understanding.

Aging is like many components of a machine falling apart at the same time. Rejuvenation will likely be made of multiple attacks. Perhaps cleaning up bad stuff from blood, like in this article, addition of new hormones or other signalling molecules, organ transplants, brain augmentation with stem cells and computer implants, etc. Not sure how teleomeres could be regrown though.

Ingest several trillion words, run training for zillions of supercomputer hours, and get back a semi-clever pointless statement about the best car to buy. It is nice that they push the envelope on how much data a model can ingest, but not sure they got anything out of it.

The desktop computer won't fade any time soon. For serious work a desktop machine is still way better than a tablet. (Granted, many times even on the desktop one works with the cloud which is server-based.) Not sure about the gain of having Windows open though. Having open standards, libraries, languages, etc., seems more important than the OS nowadays.

Back in 2001 GPL ruled. A cancer that is, even if it prevented the fragmentation of the Linux kernel. Apache II is the right license. It gives the user maximal freedom and one is not forced to release an entire project under GPL because one of your 3rd party libraries is GPL.

Speaking of the bigger picture, companies realized that it is much cheaper for everybody if the core infrastructure software is open and shared. Having things like Linux, ROS, various libraries and toolkits in the open makes it easier to build products. Especially if those products are shared as services or if the profit comes from running a cloud.

Physical books getting resold and an occasional ebook being shared or pirated causes much less of profit loss than by default allowing a relaxed policy of people getting free stuff.

Wider availability of free ebooks will mean less income for writers, obviously.