Opinions From the Grandstands

THROWING ROCKS AT SUCCESS • Part 2 of 5

Nov 12, 2026

Opinions From the Grandstand

THROWING ROCKS AT SUCCESS • Part 2 of 5

The Loudest Seats in the House

Herbert Roberts, P.E.

The test stand has room for a few people with headsets. The grandstand holds everybody else, and the grandstand is always louder.

In October, Kara Swisher took a swing at the new Tesla Roadster demo. We have evidence, because she put it on video.

The clip is a TikTok from her show, Off with Kara Swisher, posted by @onwithkaraswisher. It is a talking-head shot. Swisher wears headphones, and behind her the screen cycles through images of the Roadster, the Robovan, and Musk himself. Here is what she says, word for word:

“Elon’s car did not fly… It had a lot of suction, a lot of air compression, and a lot of flim-flammery.” She calls it “a stunt in order to focus on the fact that he hasn’t delivered the Roadster in many, many, many, many, many years.” She calls it “purely a stunt to try to get the stock to go up.” Then she widens the swing: “Just remember most of what Elon does these days is a stunt.” And she signs off: “I appreciate the dreaming, but I understand a trick when I see it.” The clip ends with a card that reads “New clips every weekend.”

Swisher deserves a fair introduction. She has covered Silicon Valley for more than three decades. She has known Musk since the late 1990s and interviewed him many times. The friendship broke in the fall of 2022. On October 17, Musk sent her a late-night email with a crude three-word verdict on her character in the subject line, over a tweet she says was actually supporting him. A month later she told NPR she had “always defended him” and still called him “amazing in rockets and cars.” She has written about the break as disappointment. Since then the coverage has run one direction.

I tell you this because her clip is the cleanest example of something new. A verdict on someone’s motives used to stay in the back row. Now it gets a microphone, a headset, a branded card, and a weekend posting schedule. Swisher has earned her seat, and she knows the industry well. That is what makes the clip worth studying: even a seasoned reporter now delivers a verdict on motive out loud and on camera, as a product. The people who build things draw a crowd, and part of every crowd comes to boo.

## The test stand and the grandstand

I had the honor of watching the engines for a fifth-generation fighter come to life on test stands, years before the public ever saw the F-22. They were built to push a fighter past Mach 2 and to supercruise, holding supersonic speed without the afterburner. No fighter engine had been built from the start to do that. Twenty years later, it is boring technology.

Those engines roared on stands a few hundred yards from the ones that once held J58s in their grip, before the J58s were bolted into SR-71s and sent off to the heavens. The J58s were on their way out. Supercruise was taking their place, for a redefined mission.

That is where I learned to love the test stand. It is a steel frame bolted to concrete, wired with sensors, and surrounded by a few people with headsets who know exactly what the part is supposed to do. When it shakes, they know why. When it fails, they know where to cut.

Behind the glass there is always a second group. Managers, visitors, and folks from down the hall who wandered over because they heard something loud was happening.

Public life works the same way. The builder stands at the test stand. Commentators, professors, and politicians sit in the grandstand. The grandstand has better seats, more microphones, and zero parts on the line.

Notice what Swisher’s clip measures. It measures how the demo looked, what it might do to a stock price, and what she believes is in Musk’s head. Those are grandstand readings. The test stand reads thrust, pressure, temperature, and whether the part came back in one piece.

## The measuring stick

Here is the rule I live by. You learn early in innovation that you never take the unsolicited opinion of a person you would never go to for their opinion. An uninformed opinion is no way to measure your actions. The data has the final word, and the data has never once cared who was talking.

Ask a simple question of any critic: would I go to this person for advice on the thing they are criticizing? If the answer is no, file the opinion with the weather report.

That holds even for a critic who knows the industry. Knowing the industry is not the same as knowing the hardware. Swisher can judge a company. The test stand judges the part.

## Why a demo looks like a stunt

So was the Roadster demo a stunt? From the grandstand, every prototype demo looks like a stunt. It is supposed to.

Start with what the hardware did. On October 15, at SpaceX’s test site in McGregor, Texas, the car clung upside down by suction, corkscrewed through the air on thrusters, and ran 0 to 60 in a thruster-assisted 0.81 seconds. Swisher is right that it did not fly the way an airplane flies. By that standard, biologists would have to rename the flying squirrel a rodent with loose skin. It glides. Nobody calls it a fraud. A test stand logs what it did do. Whether it becomes a car anyone can buy is a separate question, and a fair one.

Ubergizmo described the event as “programmed stunts on a prototype, not a showroom car shipping next quarter.” That is a stunt in the stunt-driver sense: a planned maneuver that shows what the hardware can do. Swisher means the other sense, a trick to move the stock. The first anyone can check against the video. The second is a guess about motive.

A prototype is a question in hardware. It asks whether the idea survives contact with the real world. The early answers are ugly, and SpaceX’s first rocket is the textbook case.

## Three failures, read from the test stand

Falcon 1 flew for the first time on March 24, 2006, from Omelek Island, a speck of coral in the Pacific. Thirty-three seconds after liftoff, an aluminum fitting on a fuel line, corroded by months in the salt air, leaked. Kerosene caught fire around the engine, and the rocket came apart. The cause traced back to a material choice that could not survive the island’s salt air, the kind of small detail that ends big programs.

Flight 2, in March 2007, did better. The first stage burned clean, and the vehicle crossed into space. Then propellant sloshing in the upper-stage tank set up a growing wobble until the engine shut down. The team redesigned the tanks, added baffles, and moved to a new, regeneratively cooled engine, the Merlin 1C.

Flight 3, in August 2008, carried that new engine. The first stage ran almost perfectly. After shutdown, leftover fuel in the engine’s cooling channels gave a faint push of residual thrust. On the Texas test stand, that push disappeared under ordinary air pressure. In the vacuum of space it was enough to shove the spent first stage back into the second stage. The stages collided, and the mission was lost.

That one humbles every engineer who reads it, me included. The test stand itself missed it, because a test stand at sea level can only tell you what sea level allows. The fix was simple once someone found it: wait a little longer before separating the stages.

Less than two months later, Flight 4 lifted off on September 28, 2008, with that longer delay built in. It reached orbit. Falcon 1 became the first privately developed liquid-fueled rocket to do it. Musk has said that if the fourth launch had failed, SpaceX would not exist.

From the grandstand, that was three stunts in a row. From the test stand, it was three findings, each one different, each one written down and fixed. Those three failures were the work.

The Roadster passed its demo. Its next test is whether it ships. A demo becomes a product the day it ships. Until then it lives in a gray zone where a fair observer withholds the verdict and an unfair one hands it down early.

## The critics’ best point

The critics have one argument with real weight. Journalism is a watchdog, and watchdogs are supposed to bark at powerful people. Missed deadlines are also real. Tesla unveiled the new Roadster in November 2017 with production planned for 2020. The date moved to 2021, then 2022, then 2023. The flying demo slipped too: end of 2025, then April 1, 2026, then late April, then August, then October 1, and finally October 15, after weather scrubbed the October date. Musk now guides production for 2027 or 2028. Swisher’s “many, many, many, many, many years” is a fair jab, and customers who put down deposits deserve the reminder.

Here is my answer. Bark at the deadline, and leave the motive alone. “This car is years late” is a fact anyone can check. “This is purely a stunt to move the stock” is a claim about what is inside a man’s head, delivered from a seat that has never built a car. Late hardware is a schedule problem. The watchdog earns trust by reporting what the hardware did and saving the mind-reading for fiction.

A skeptic will say the record of missed dates is data too, and an engineer should read it. Fair. That record supports “it will be late again.” It does not support “it is a stock pump.” One is a forecast from the schedule. The other is a guess about a man’s head.

I counted her podcast and essay titles about Musk and his companies from October 2024 to October 2026: 78 in all. I read titles only and sorted them by my own judgment of their wording. Some come from Pivot, which she co-hosts with Scott Galloway. Thirty read negative, three read positive, and 45 read neutral. A few examples, so you can judge my sorting: [ADD 2–3 NEGATIVE TITLES], [ADD 1–2 POSITIVE TITLES], [ADD 1–2 NEUTRAL TITLES]. That is ten boos for every cheer. The rockets are still flying.

## Back to the stand

The grandstand will always be louder. It has the microphones, the clips, and the comment sections. The test stand has something better. It has the people who know whether the part held.

Pick your measuring stick before the noise starts. Choose the people you would actually call for advice, and let everyone else cheer or boo as they like. The verdict that matters gets written at the test stand, on the day the part comes back in one piece.

---

*Tuesday, Part 3: Just Photoshop Me Next to Edmund Hillary and That Other Guy. On Everest, the Sherpas fix the ropes and carry the packs, and the paying clients pose at the top. Resentment grows in the gap between the climb and the picture.*

Herbert Roberts, P.E. is a licensed professional engineer with 30+ years in aviation research and development across three companies, in addition to analyzing accidents under his P.E. license.

Discussion about this post

Ready for more?