First, humans use other senses like hearing. Second, humans use reasoning. Humans are better at making inferences to fill in missing information. Humans also use world models to help make decisions. Not a whole lot of AI approaches do this. However, the machines will ultimately win. Human performance sucks. It isn't hard to beat it. We already see this in aviation where the standards and expectations for machines are much higher for political reasons even though the autopilot is the better pilot in the cockpit

Thanks to California, the threshold trim board I bought a few years back was labeled as containing cancer causing materials. Yup, wood, water-based finish, and plastic wrap.

Up until a couple of months ago, I was using a 15 year old router running OpenWRT topping out at 802.11n! It finally became too unreliable - the radios would reset very occasionally at first like once a week for the longest time then several times an hour towards the end. I know Flint 2 is just a little behind the times, but it was recommended for its OpenWRT support, and anything on the market today is light years ahead of my old router.

In the real world, the problem isn't really creating laminar flow. It's dealing with it after it forms. Laminar flows tend to be less resilient to real world conditions than turbulent flows. Imagine an aerodynamic stall that, rather than happen in a benign and predictable way, happens without warning and in unpredictable ways. Or suddenly the laminar flow went turbulent because of surface contamination but you designed the engine size based on the drag from laminar flow. Interesting science, but I'm not sure it changes engineering much for the foreseeable future.

Many more people die because of cold than heat. So who's cherry picking now?

What if I don't give a crap about carbon footprint? That's only one of many metrics that could be used. And it happens to support your position.

PINs don't have to be 4 digits. I use a smart card for login at work, and the PINs are required to be at least 8 digits long. The system 2-factor. The PIN is a something I know. The smart card is something I have. The PIN essentially verifies the smart card, and then the certificates on the smart card, which are much higher security, are used for everything else.

If you want heated seats, it isn't that hard to get power to them via a switch that you install. In some cases, it may be possible to use the factory switch if it's sending signals over CAN or another network.

Or pull the fuse to the telematics module. I've seen several car repair videos where they do that to stop a large parasitic draw when the module goes bad. It will set a permanent code, but it isn't the type of code that causes the idiot lights to come on.

But the metadata isn't encrypted afaik. That's often more revealing than the message content. For example, if all I have is sender and receiver, I can start building relationship maps or org charts. The number of interactions between sender and receiver can indicate the importance of the relationship. If I have time stamps, I can start building a pattern of life.

Look up eLORAN. This has been studied thoroughly before, and the conclusion was develop eLORAN. The Coast Guard is going to have a heart attack though.

There's a new version called eLORAN. It meets Stratum I timing requirements like GPS but is terrestrial based, high power, and low frequency. Basically the opposite of GNSS. eLORAN stations can be treated like pseudolites due to the precision timing, and instead of doing time difference for hyperbolic navigation it's very similar to pseudorange calculations like GNSS. So you don't need LORAN chains like before.

This has been looked at extensively before. The conclusion was to develop an eLORAN system. But someone in DoT had a stick up their butt and kept killing it. I think Collins even developed prototype avionics tying eLORAN, GNSS, and INS together. eLORAN even meets Stratum I timing requirements like GNSS. Interesting bit of trivia. Before the Coast Guard decommissioned LORAN-C, cell towers had LORAN-C receivers as a timing backup to GNSS

The article mentions the need for data manipulation languages. Wasn't COBOL designed for this? We keep saying COBOL needs to die, but does it? Maybe it just needs to modernize.

This used to happen. But the Democrats decided they really wanted the 16th Amendment. You've got to lie in the bed you make.