The projected deaths for Covid-19 in the USA are horrific -- 100,000 to 240,000. Let's hope it's not nearly that bad, but those numbers are enough that they actually could alter the election. Not simply because Covid-19 will be the top issue in the election, but because voters will die.

Recently, Stefan Seltz-Axmacher, the founder of Starsky Robotics -- a startup doing self-driving and remove-driven transport trucks that I advised before they started going -- wrote a detailed and complex blog post about why he feels his company had to shut down. He goes into several issues, including failures of Deep Learning to meet hype, VC desires, strangeness of the trucking industry and lack of love for safety.

In my new article for the Forbes site, I dig into those reasons and whether he's right that nobody else will succeed soon, either.

People are doing huge amounts of videoconferencing during the Covid crisis. The tools keep improving, but there's a great deal that individual participants can do to make the meetings better. They take some effort but it's worth it.

In some discussion, I have seen it become almost an assumption that the economic meltdown and the Covid crisis will erode confidence in the President and settle the election, presuming things continue to November as they likely will. Historical patterns suggest that Presidents with good economies and stock markets get elected, those without them don't. We're seeing economic meltdown, high unemployment, fear and more.

I've been involved with delivery robots for a long time, and on my walk through empty streets yesterday, I noticed a certain irony. We have a desperate need for more delivery capacity, especially without humans handling packages, and teams have been working hard to make deliverbots safe enough to drive on our streets.

It is possible if, among the swing states, Trump wins Florida, Pennsylvania, North Carolina and the Nebraska 2nd (Omaha), while Biden wins Michigan, Arizona and Wisconsin. This is not one of the most likely scenarios, because Arizona and Wisconsin are currently more on Trump's side than Biden's, but it's possible.

If we shut down public areas, we're going to need a lot of online shopping and home delivery. How can we do that in a virus-infected world? Here's some plans for how to make it happen even with gig workers (who aren't driving Uber and Lyft much any more.)

I outline some of the ways to make it work in this Forbes.com article.

As LAX and other airports push ride-hail to remote lots (which you have to take a shuttle to in the case of LAX-it) I examine why that's a crazy decision in my new article at Forbes.com. In the article I also touch on how we can eventually move to being picked up, not at the curb, but at the plane, in an airport with lots of robocar pods.

LAX won't let Uber pick you up at the curb. It should be at your plane

I have built a spreadsheet summarizing the recent poll results in head to head polls of a Democratic candidate and President Trump. These are the only polls that matter. Do not look at or cite national polls. Results and pollster quality ratings come via 538.

An EasyMile made a sudden stop from 7mph and a seated passenger fell off her seat to minor injuries. Now NHTSA has ordered EasyMile to stop testing with passengers.

Transit shuttles don't usually have seatbelts, but maybe EasyMile needs them during the testing phase. But can it ever take them out?

This weekend I went to the finals of the GoFly prize, a Boeing sponsored contest for personal VTOL flying machines. Sadly, nobody was able to build one that could meet all the requirements in the rules, and only a few of the contestants could even fly. That was disappointing, but then so was the first Darpa Grand Challenge.

The California robocar disengagement reports are out. And everybody is now pointing out that they're not very useful because everybody uses different methods. So I have an article about what we do learn from the data, little as it is.

Read California Disengagement Reports aren't too engaging at Forbes.com

Yesterday's NTSB hearings contained much stronger criticism of Tesla Autopilot than ever before. In particular the NTSB made recommendations:

Many of the media were keen to pick up on a report from McAfee researchers about how they were able to simply modify a speed limit sign to cause the MobilEye in old Teslas to misread it and speed up. We get spooked when AI software acts like an idiot. But in reality, this isn't the sort of attack that is likely to be done in the wild, and it's also unlikely to cause any danger.