Are you referring to the Apple Orchard? I know most of the sites that had manufacturing fall into the toxic category, but my fuzzy memory is that you worked around there.

I have done work in some former IBM sites, and even the very old ones can be quite interesting. (As an architecture nerd I would enjoy exploring an IM Pei building though.) Only problem is they are abandoned because of hazmat and ground remediation is a ~75-year process.

Presumably it can help fill undesirable shifts by offering higher pay, but it would seem like you need a huge pool of people to make it work.

Are you kidding with the Lane Assist for 10 or 20 years? Lane assist from 10 years ago? Tesla was the only one that had a functional product for lane centering in 2016, but that is nothing compared to what the autonomous systems can do today. The Japanese cars I have driven recently have pretty functional adaptive cruise control / lane centering, easily on par with Tesla circa 2019, but things have moved forward.

I drive a 2020 Model Y and while it is on the old FSD stack still, surface street driving is just fine. (Highway driving on 4+ lane freeways is miserable though.) It does actually pull over for emergency vehicles (but not reliably enough that I would want to leave it in FSD when I see one coming). It gives way for bicycles by crossing into the oncoming lane if there is time. It will follow the car in front of it through complicated construction zones (but I take control if I am the lead car). It has quirks, but I would score it better than most 18-year old or 80-year old drivers.

"Perfect" is impossible for AVs or humans alike, and when you mix the two you are bound to have some screw ups.

That said, Waymo clearly needs to do better. Tesla seems to keep its number of autonomous vehicles/miles down in order to stay out of the news.

Unfortunately for Samsung the chip business is going to cause a lot of pain in all the other divisions. Apparently phones, tvs, and white goods are all dramatically reducing product offerings, staffing, and general competitiveness.

They have to paint a pretty picture or the backlash will be too restrictive for their goals. The problem is the technology is a long way off and they need a lot of "concessions" to buy that time.

Input energy 6MWh, output energy 4.5MWh: 75% efficiency. It is disingenuous for a manufacturer to state battery chemical round trip efficiency when the user is buying a complete package.

The battery is set with reserves for about 24h of backup, so essentially I have one module for arbitrage and two for backup energy. The parasitic losses account for ~12% of my capacity per day. I have enough PV that it isn't a problem, but for most people it will lead to disappointment.

I installed a 45kWh FranklinWH system and it has parasitic losses of 1.5MWh/year on total discharge energy of 4.5MWh. Essentially each module you add needs another 400W panel to cover losses. The advertised 97% round-trip efficiency is a joke.

Emergency response window for major disaster is usually around 4-8 hours for initial mobilization. Stabilization takes about 24 hours. If backup power is still needed then deploying a generator or refueling in that timeframe is feasible. Lack of effective turns an emergency into chaos and emergency response teams are usually the "second responders"-- people on the ground need to be effective without special equipment.

24 Hours is better of course, but for a "small" 5G tower that is about 125 kWh and a full size tower twice that. That is about 80 gallons of diesel or 1.5 tons of batteries. The way the 5G networks are set up it is a big ask, even if it is needed as emergency infrastructure.

A minimum of about 8 hours is more realistic for cell towers and 72 hours for core network is what you need for reliability and recovery. 24 hours for the towers is better, but difficult to achieve for many sites.

They also fail to understand that the problem is continuous. As you have changes in materials, methods, regulations, etc. you need to re-train the systems to apply new experience so the problem is never ending. You can save for a few quarters, but long term you are still killing yourself.

But then you need about 5 acres for support equipment and another 40-100 acres for a security perimeter (minimum). It is also just 1/3 the capacity needed for current generation data centers.

A simple solution for the Roku example is to use HomeAssistant or similar to connect and control.

The basics are pretty easy to guide someone through, but the maintaining and dealing with tuning what traffic can go to trusted networks gets pretty complicated (especially with IPv6, at least for me). I have a few lazy cop-outs (HomeAssistant server has two interfaces (IoT and LAN) rather than being in its own VLAN with traffic rules to address specific ports and applications. (At one point I had a server VLAN that it sat in, but it has different needs than Influx or the PiHole or the NAS.)

Proper security requires vigilance and that is much harder to train.