Poke further at the story. Look at the battery charge graph, and more specifically look at its slope, which is the fuel efficiency. Shifting to cruise control did not affect the drainage curve at all, and when he called Tesla they suggested that cruise control was a Bad Idea, because the Tesla has regenerative braking, so you see that he turns it off at about 225 mi and starts stop-and-go driving in the city. This devastates the Tesla's fuel efficiency. It does this for a very simple reason which should be high-school physics: your brakes are energy dissipaters and are bound by the Second Law of Thermodynamics. Regenerative braking just lowers the loss -- it does not eliminate it.
Now, if that advice was given, then the journalist trusted Tesla when they said something which contradicted high school physics. That is a dark smear on Broder -- at least if Broder was planning to be a scientific journalist -- but also a dark smear on Tesla's customer support. What were they thinking? But the two accounts are immediately reconcilable now.
Broder now is thinking, "okay, stop-and-go city traffic will use the brakes and recharge the car, this city driving will have a less negative slope." From this impression he probably expected he was at 20% charge when he stopped; he was actually at around 8 or 9% and it immediately drops 1% overnight. That should chop 1/10th off of his remaining miles, but he claims it went from 90 miles to 45.
So, here's the story: He is expecting a number at around 100 miles because he thinks he's being more efficient now. He sees 50 miles. On a digital display at a hotel at night, he misreads the 5 as a 9, and this fits with his expectation, so he goes to bed thinking he has 90 miles of range. He wakes up and the Tesla has lost 5 miles of range inexplicably -- but it therefore has become 45 miles, which looks totally different. He calls them up complaining that the Tesla lost half of its charge overnight and some sympathetic tech support describes it as a "software glitch."
That's a perfect storm scenario right there, because now he thinks that he does have the extra range and that braking is good for the car and that the car is simply misreporting what it can do. Confirming this, he makes it to the Milford supercharge with less than 0 miles of range, and charges it back up again to 185 miles. He is confident now, and you see him averaging 65. There are a bunch of full stops near 400 miles, but remember, he thinks that full stops are good. With a bunch of this, he stops and calls up Tesla. "What the hell, I can't get back to Milford on this expected range, can you find me a nearby charger?" He goes 11 miles in the opposite direction, plugs into a station in Norwich for an hour, and he visits a diner. He confirms that he only got to 28% and really should have let the car charge here because the display never got as high as it should be, but he says that Tesla had cleared him to go to Milford and that he trusted the humans at this point over the sensors in the car.
He gets back on the highway at 45/50 to try to conserve power (which does not affect the slope of the curve all that much) and that is the subject of this second-to-last Google Map. The sensors are right, the tech support was misguided, the car stops and needs to be towed. Both stories fit pretty well with each other.