I think the answer is no if you don't have clean access to the proprietary software, e.g., if you decompile or reverse engineer it in violation of a license agreement that you agreed to. That taints the spec, which taints the clean room reimplementation. I think this also applies to leaked software - if you know it's someone's trade secret, but you use it anyway to create a competing product, you can be sued.

But then if you want to plan meetings with people in other regions, you have to keep a table of "customary business hours" for every part of the world for every time of year. That ends up looking like a time zone / DST table, but without the standardization.

Theoretical and prophylactic are not the same thing. Quantum decryption is indeed still a future threat, but that doesn't mean you shouldn't protect yourself now. Various parties are already using a technique called "harvest now, decrypt later" to record web traffic today that they may be able to decrypt and read with quantum computers later. So using non-quantum-resistant encryption today is as risky as using plain text, just with a delay before the consequences appear.

I've ridden in one. It was just like riding in any car, but a little disconcerting to have the driver's seat empty while the car did its thing. Maybe a little smoother and more predictable than the average Uber.

Given how many Waymos there are on the road and how little trouble weâ(TM)ve heard about - despite breathless coverage of every incident, - Iâ(TM)d say they seem to be doing better than human drivers.

This is a fallacy. To halt climate change, we don't need to move away from extracting hyrdrocarbons, just from burning them. It would be a huge step forward to convert hydrocarbons to plastics using renewable electricity, then eventually recycle or bury those plastics. Using fossil hydrocarbons as a feedstock is potentially a zero-emission process, unlike using them as fuels.

I love the idea of e-fuels. But it would be cheaper to replace most of those cars with new EVs than to fill them with e-gasoline at $0.60 (extra) per mile, i.e., $600/mo. for the average car.

At the quoted costs, e-gasoline is about $1600/tCO2 avoided (say $15 extra per gallon of gasoline, which avoids 9 kg of CO2 emissions). Unfortunately, thatâ(TM)s around 8 times higher than the harm done by CO2, so even serious environmental economists would say itâ(TM)s not worth using unless costs can be brought way down.

As a starting point, consider using just the CO2 part of this system to sequester CO2 from aircraft. Or switch the grid to 95% renewable power and use the CO2 collector to sequester the remaining CO2 emissions.

Iâ(TM)m so tired of tech people coming up with new, unnecessary, overpriced solutions to renewable power variability or car range anxiety. Just stick in some batteries! This thing costs $15,000-$20,000, which means a $300/mo payment to get 30 gals of gasoline. Thatâ(TM)s $10/gal just for the equipment. Then theyâ(TM)re using 75 kWh of electricity per gallon, which costs around $30 retail or $10 from a home solar system. All to get you about 9 kWh of propulsion (after the 75% losses in the car engine). You could replace all of this for $1.35-$4.00/gal equivalent with an EV.

The summary says

This implies that in a rosier world, a majority of the companies would be optimistic about revenue growth. But raising revenue for most or all of the companies is not possible unless the customers (a.k.a. all the companies' employees) have more to spend, and no one is talking about raising wages as a result of AI. So where is all the extra revenue supposed to come from?

Time from bug creation to fix and time from bug discovery to fix are both important. But only one is easy to measure, and thatâ(TM)s what the authors measured. This blog post wasnâ(TM)t driven by âoewhat question is important to answer?â, it was driven by âoewhat can we say with these data?â.

Interesting point. To make time zones work with metric time, we'd also want to switch to metric angles for longitude. Each metric hour (1/10 day) would correspond to 40 gradians or 0.2 pi radians.

One advantage of this: we'd only need 10 base time zones, which might be easier to keep track of than 24. But people on the edges of the zones would have pretty asymmetric sunrise/sunset times.

I was ready to say "yay, weâ(TM)ve finally reached the point where AIs can find code and network vulnerabilities and we can patch them and stop worrying so much about exploits." But then I realized, on the other hand, it might just be a battle of AI vs AI now: can the white hat AIs find the exploits faster than the black hat AIs? I wonder which will win out?

Put another way, maybe there's an infinite supply of exploits (especially with vibed code) and the bad AIs may be able to find them faster than the good AIs can eliminate them. Or maybe there's a finite supply of bugs in any system and once they're eliminated it will be safe and secure.

TFS says "The company's fundamentals improved dramatically over that period," but then uses 1999 as the baseline instead of the peak day in 2000. I think thatâ(TM)s how it's possible that "profits have quadrupled" while "earnings per share have grown eightfold": the article skipped over the final 2-for-1 split in Sept 1999. Pretty poor financial reporting though.

I spent a while as a professor of electrical engineering. As a student, I always hated when tests were a time trial to get as much done as you could in the time available. Yes, being prepared helps with that. But those tests evaluate calmness under pressure, speed of writing and effectiveness of test-taking strategies more than the material you learned in the course. And if there's only time to answer half the questions, then students only need to know half the material to get a decent grade after the curve. (I also hated mis-written test questions where there was no correct answer.) Speed tests don't predict real-world performance either, because at work you generally have enough time to think (and references), but you have to get it right.

So as a professor I always sat down and answered (and timed) every question on my tests in advance. Then I trimmed it down until there were only 17 mins of material for a 50 min exam. That eliminated speed as a major factor. So then there was also no advantage in taking extra time. If you knew the material well on test day, you would do fine, and if you didn't, you wouldn't. I wish more professors would take this approach.

OK, but how did you add them together? My wife added the 9 and the 8 to get 17(0), then added the 8 and the 7 to get 15, then carried the extra 1 back to the 10s place to get 185. So basically what they taught us to do on paper in the 1970s and â80s, but working from left to right instead of right to left (which I also often do, and is probably some other named shortcut).

But I kind of smooshed 98 and 87 together in my head and visualized the 87 giving 2 to pad out the 98 to 100 and leave another 85. That is convoluted to explain but natural to do if thatâ(TM)s how your head works. Which is exactly my point â" no need to teach these shortcuts!