Featured Post

This is the Kodak Moment for the Auto Industry

Plug-In Drivers Not Missin' the Piston Electric vehicles are here to stay. Their market acceptance is currently small but growing...

Tuesday, March 30, 2021

Musk's Empire - The Power of Engineering



Science is the body of knowledge learned from exploring the physical and natural world. 
Engineering is the application of knowledge in order to solve problems and fulfill needs.

No offense to Andy Weir but Mark Watney was applying his knowledge, so he didn't "Science the 💩 out of this," rather he "Engineered the excrement out of it." 😃

If you ask a scientist about a bridge design, they might say, "It works under one model, but fails under another model." So if you want to know if it would work in the real world, ask an engineer.

Engineering is about getting things to work in the real world. This is very different than getting something to work in a controlled lab. The real world has to contend with massive temperature variations, dirt, grime, margins of error, tolerances, weather, metal fatigue, wear and tear, etc.

Science's job is to be out ahead of the rest of us, making discoveries. The applications of those discoveries may not be readily apparent. When Michael Faraday first discovered the electro-motive force, his initial demonstrations were not much more than a metal needle spinning in a conductive fluid. At one of his demonstrations, he was asked what use could this invention ever have. His response was, "What use is a newborn baby?" Meaning this discovery had much-unrealized potential. Faraday didn't stop at just discovering the science; he went on to do the engineering work too and made the first electric motor and soon after the first electric generator. His discovery and the inventions that followed are at the heart of our modern world. Nearly all electricity (except solar) is created by turning a generator. This discovery was crucial; however, it was the engineering work that converted the scientific foundation from a newborn babe into useful machines.

Applying This To Elon Musk

One criticism of Musk is that the things that he's delivering were not his ideas. Musk is known for SpaceX, Tesla, The Boring Co., Starlink...  The criticism goes something like: there were electric cars long before Musk was born, they were not his idea. The "vac-train" was invented in 1799, so Musk's Hyperloop idea was not a new one. Starlink was not a new idea, SpaceX had even launched satellites for OneWeb. They took the idea from their customer. Landing rockets has been common in sci-fi since Buck Rogers in the 1930s and Bell Aerosystems demonstrated vertical landing rockets in 1961. Musk and Co have never had a unique idea. </crit>

Let's disassemble this criticism. I have an idea for a matter transporter. No wait, that was Gene Roddenberry. I have an idea for a time machine. No wait, that was H.G. Wells. Now if I actually built one of these (or both, because why not), would you say, "You just copied someone else's idea!"? Of course not. Imagining these is not the hard part; creating them is where the difficulty lies.

The idea is important, but without the engineering work to bring it forth into the world as a real thing that we can interact with, it's just an idea. Great for fictional works or demonstrations of a needle spinning in a dish of water, but these alone will not bring about the next industrial revolution or change how you and I live our lives.

Musk's accomplishments (actually all the engineers and technicians that work for him) are not coming up with new ideas. The accomplishments are solving all the problems that were encountered going from concept to product.

There are millions of ideas out there. What matters is selecting from them, something that is possible, meets a need, and has not been done yet. It's about looking at the good ideas that we already have and then making them work. It's about raising the newborn babe into a functional adult.

Let's deep dive into one specific product: Solar Glass Roof. The idea for roofing materials with integrated solar is not a new one. There are several patents (including one from NASA) dating back to the 1970s for "building-integrated photovoltaic."  Many many businesses have tried to make this into a viable product and failed. Some were solar roof tiles, some had flexible roll-on solar that went over a metal roofing, others even had solar collecting windows. In 2016, Green Tech Media wrote an open letter to Elon Musk and the Brothers Rive (then running SolarCity) asking them (nah, pleading with them) not to go forward with a solar roof product. The author didn't want SolarCity (this was before they became a part of Tesla) to die upon that precipice as had so many other solar companies. The author lists nearly two dozen companies and solar roof products that seemed like great ideas but never made it beyond the prototype phase and in some cases, took the company down with them. 

Musk didn't heed this warning; he couldn't be waved off. He uses first-principle thinking to determine what is and what is not possible. This thinking had led him to determine that a solar glass roof was possible. Musk also adheres to the motto, although it may be difficult, that doesn't mean it shouldn't be attempted. He has said that when Tesla started, he knew the most likely outcome was failure. 

Tesla moved ahead with solar roof developments. They ran into many of the same problems as the other companies. However, Tesla had money, time, and brilliant engineers. Previous attempts were missing one or more of these vital ingredients. Tesla did not make the bulk of their revenue from solar, so they could take their time and work out the bugs. They had problems with wiring, couplings, clips, installation time, and more. They iterated the product, fixing, refining, improving. They hired roofing crews and had them install solar roofs on test homes and then looked at the results (how long did it take, did they get it right, how much breakage...) and they iterated more. Finally, in the second half of 2020 (five years after the Green Tech Media open letter), Tesla started installing solar glass roofs in quantity. They now have several roofing companies certified to install these shinny black roofs all around the US and they are expanding into Canada in the first half of 2021. 

This is a perfect example of the need for great engineering. The science of solar roofing was solved. Integrate a solar cell under a transparent protective roofing tile, nail them down, hook them up, and easy-peazy you have a solar roof. However, to make a product that can be installed in a day, will last 20 years under the Arizona sun, and survive Canadian winters is a difficult engineering problem.

The Physics of The Impossible

In his book, Physics of the Impossible, Michio Kaku defined 3 levels of Impossible ideas: 

Class I impossibilities - technologies we have no idea how to achieve today, but they do not violate the known laws of physics, e.g., energy shields or artificial gravity. These are things that might be achievable in a few hundred years plus or minus (mostly plus). 

Class II impossibilities - technologies that "sit at the very edge of our understanding of the physical world." These (like Class I) do not violate the known physical laws, but we know a lot less in these areas. We need to learn much more to truly understand the feasibility and, if they are deemed possible, they may require the entire energy of a star or a black hole. That certainly makes them impossible today, but with a thousand (or millions) of years of advancement may become possible. 

Class III impossibilities - These are the things that (based on the science we already know) are impossible. For example, perpetual motion machines and precognition are class 3 impossibilities.  

Musk and The Physics of The Possible

Why did we go through the list of impossibility categories? Because the first step to creating an innovative product is knowing what's possible. 

When one of Musk's companies wants to make something, they don't ask if something has been done before or how it is currently being done by their competitors. The first question is, "Is it possible? If physical atoms are arranged properly, can they function as needed?" This "atoms up" thinking, means they will try things that been attempted before either with the ignorance of past failures or hubris based on the list of "previously impossible" things that Musk Co. have accomplished. This is what I call a Class ½ Impossibility. There are no laws of physics that limit its possibility, you don't have to wait 100 years for it, but it won't be easy; perhaps some prototypes or lab samples have been created, but it has not been productized due to some obstacles. 

It's relatively easy to take an existing product and make a derivative or a minor improvement. However, this will not result in a breakthrough product. If you want to productize something that's never existed beyond the prototype phase, that will be far more difficult. You might even say it will be hell, but as we said in our Mistaking A Clear View For a Short Distance article, Musk knows it will be hell; he's familiar with hell, he's walked the trails through Hell so many times, he could be a tour guide.

Musk has been on the trail through Hell so many times that he could be a tour guide.

Tesla's Innovation Moat 

This tendency to take on Class ½ Impossibilities gives Tesla (and other Musk-o-verse companies) a competitive advantage. Most businesses are worried about next quarter's results and would not be willing to fund product development for years without assurances that it will be profitable. 

Full Self Driving is another example. This falls into the category of things that are possible, but not yet productized. Tesla has been working on this since at least 2015. They have hired a hardware and software development team with some of the best talent in the world. According to our estimate, Tesla will be working on this until 2027 before they have a true Level 5 robotaxi fleet. That's 12+ years of engineering work before arriving at the destination. Although this estimate is a decade behind Musk's original estimate, it is still likely 5 years ahead of competitors. That's more like a Silicon Valley tech company than a car company. That's Tesla's Innovation Moat.


Disclosure: I'm long TSLA

Monday, March 22, 2021

Spring 2021 Has Sprung! First Net Positive Day w/ Tesla Solar and Powerwall


On Friday March 12th, we generated more electricity than we used. Here in NW Oregon, we usually have short cloudy/rainy days in the winter. Well, spring is here and the days alternate between sunny blue skies and rainy.

When July/August hit, we'll have more sun, but the AC will be on, boosting our energy use, raising the bar for energy-positive days, so these sunny spring days are the easiest. 

Looking at the timeline for our net positive day, there are a few interesting moments. 
  • 3AM the EVs charge up. You can see the big spike in energy use. On a power level, this event is far bigger than anything else that happens throughout the day. 
  • 6AM peak rate hour start. The Powerwalls start discharging to remove our home from the grid load. 
  • 7AM the sun comes out. Our PV system started generating power. Since we were running in Cost Savings mode, the PV energy is sent to the grid during peak hours. 
  • 10AM peak ends. I'm not sure why, but the battery continued to discharge. This is not what usually happens. 
  • Noon Powerwalls start charging with solar. 
  • 1PM Powerwalls stop charging. They were not full but stopped charging for some reason. 
  • 2:30 PM Powerwalls start charging again. 
  • 5PM evening peak starts. Once again the Powerwalls start discharging to remove our home from the grid load. 
  • 8PM peak ends. The evening peak ended and yet the Powerwalls continued to discharge for another 2 hours. 
  • 10PM off-peak. As we hit off-peak, the Powerwalls stopped discharging. 
Why did the Powerwalls continue to discharge? We have the system in Advanced mode which is an intelligent mode that uses weather forecasts to "make room" to store anticipated solar production. That means that occasionally you might see mid-peak discharging. 

Of course, if you don't want all of this charging and discharging, you can just put the system in Backup-Only mode, but that's no fun.

Happy Spring Time, enjoy it while you can.  

Monday, March 15, 2021

2020 Solar Production Recap

We had solar panels installed on our home in late 2007. That makes 2020 our 13th full year with solar panels on our roof. That's over 13 years quietly, dependably, making electricity to run our home. We started with a 4 kW system and then added another 8 kW in 2015.

In 2020, together these two systems generated 12,345 kWh. Yes, it was really 1-2-3-4-5. This brings our lifetime total to 94,137 kWh. At 14¢ that would be ~$13,000 worth of electricity. If these 94 MWhs were used to charge a Tesla Model 3 Standard Range Plus, it would be able to drive more than 600,000 km or nearly 16 times around Earth. 

Here's a chart of our lifetime production: 


If you look closely at the summer of 2020, you can see that the production flattens out faster than in previous years. That was due to the various fires that darkened our skies. It's ironic that the fires are made worse by global warming and they are reducing our solar production. A vicious feedback cycle.

2020 ended with our Powerwall installation on New Year's Eve. 

Referral code

If you are interested in a Tesla solar glass roof or their solar subscription program, you can use my referral link and after activation, you'll get $100 of free solar energy. 

Monday, March 1, 2021

When Will Tesla Cars Drive Themselves?


Tesla vehicles have amazing technology. They have big screens, real-time traffic, over-the-air updates, streaming music, you can play video games, and even watch Netflix movies (while parked). 

More to the point of this blog entry, they also have Autopilot. Autopilot is an advanced driver assistance system (ADAS) that includes:
  • Traffic-Aware Cruise Control: Matches the speed of your car to that of the surrounding traffic
  • Autosteer: Assists in steering within a clearly marked lane, and uses traffic-aware cruise control
As an ADAS, it is clear that the driver is required to be alert. This system is intended to ease cognitive load and allow the driver to have better situational awareness. It is *not* intended to allow drivers to use their phones or take naps. 

The next level up from Autopilot is Full Self-Driving (FSD) Capability. This includes: 
  • Navigate on Autopilot (Beta): Actively guides your car from a highway’s on-ramp to off-ramp, including suggesting lane changes, navigating interchanges, automatically engaging the turn signal, and taking the exit
  • Auto Lane Change: Assists in moving to an adjacent lane on the highway when Autosteer is engaged
  • Autopark: Helps automatically parallel or perpendicular park your car, with a single touch
  • Summon: Moves your car in and out of a tight space using the mobile app or key
  • Smart Summon: Your car will navigate more complex environments and parking spaces, maneuvering around objects as necessary to come find you in a parking lot.
  • Traffic and Stop Sign Control (Beta): Identifies stop signs and traffic lights and automatically slows your car to a stop on approach, with your active supervision
  • Autosteer on city streets (coming soon as I write this)
There are a few important things to point out about "FSD Capability." The first thing to note that all of these features still require driver supervision and many of them are still in beta or even in limited release beta. That is to say, this is not yet a fully autonomous vehicle, despite the FSD monicker. The second thing to point out is the word "capability." A piece of paper is capable of being folded into a paper airplane, but that does not make a flat sheet of paper an airplane. So when you buy the FSD Capability, you are buying a sheet of paper with some folds in it, and the promise that there will be updates to finish folding it the rest of the way.

If that's what you want to buy, good on you, go for it. I did and I've been very happy with the progress of the FSD Capability. I just wanted to paint a clear picture that when you buy the FSD Capability, you are (currently) still getting an ADAS, just one with more functionality than most. 

Musk's "Coming Soon" Comments 

When will Tesla's FSD evolve from 'just' ADAS to a true fully autonomous vehicle? Elon Musk has been asked this question in several public forums. Here are some of his replies:

[12.22.2015] “We’re going to end up with complete autonomy, and I think we will have complete autonomy in approximately two years.”

[10.19.2016] Discussing the new Autopilot hardware, Musk said, "It is fully capable of Level 5 autonomy, a big step forward." *Note the work 'capable' verses something like 'functional.'

[06.12.2018] Tesla’s cars will in August suddenly activate “full self-driving features,” the company's chief executive Elon Musk tweeted on Sunday. "Features" meaning some subset of "Full."

[10.21.2019] "Next year for sure, we will have over a million robotaxis on the road"

[07.09.2020] "I remain confident that we will have the basic functionality for level five autonomy complete this year." Where "basic functionallity" is again a subset of "full".

[01.28.2021] "Basically, I'm highly confident the car will drive itself with reliability in excess of a human this year. ...we need to probably do a little bit more work to prove that Tesla Autopilot is capable of full self-driving, which, I think, will become obvious later this year."

As you can see, in statement after statement, Musk had high confidence that FSD-Level 5 will happen soon (where "soon" is any time 2017 or later). I submit to you that all forward-looking statements about something that's never been done before should be taken with a grain of salt. Additionally, these statements are often made under a Safe Habor Clause that provides legal protection for optimistic speculation. Careful parsing allows you to find the caveats (e.g., 'I think', capable, features, basic...), that a quick enthusiastic read (or an intentionally malicious misread) will miss. If you want a vehicle to drive you from A to B tomorrow, I suggest that you take the train, a bus, use Lyft, or hire a chauffeur rather than buy FSD.

If you want a car that can drive you around right now, while you look at your phone, FSD is not it. Take a Lyft instead. 

If, however, you want to be involved in testing one of the most exciting robotic systems that humankind has ever created (and you don't mind continuing to be a responsible driver), then FSD might be worth your hard-earned money. Today, using the Smart Summon feature a Tesla vehicle can (usually) pull out of a parking space and drive to you in a parking lot. This is a nice feature to have if your parking spot has flooded or you have large/heavy items that you don't want to lug through the parking lot.

If you want to test one of the most exciting neural net robotics systems that humankind has ever created (while continuing to be an attentive driver), then FSD is worth the money.

So When Will Full Autonomy Arrive?

In other (less optimistic) statements, Musk has said that they'll have to "chase the nines" to get to full autonomy. I think these are the more realistic statements from Musk that acknowledge that there are many many edge-cases that they will need to resolve and that will be a long process. Musk is a perennial optimist and I don't think his estimates are meant to be deceptive. Rather, they are ambitious goal statements that he believes to be possible. Musk excites, motivates, and manages using moonshots and urgency. It is a proven formula at Tesla (and SpaceX). They have accomplished things that no other company has ever achieved (just not on the publicized timeline). It is also important to note that even though Musk's initial target was 2017 and full autonomy has still not arrived, Tesla has continued to make forward strides every year. Sometimes, when you are doing something for the first time, there are challenges that you didn't even know existed until you get there and encounter them (unknown unknowns). 

Stop Sidestepping the Question! When Will It Happen? Give Me A Date!

It is impossible to predict. Neural nets have a strange tendency to be incompetent for many many iterations and then, suddenly, something clicks and they do amazing things far surpassing your expectations. This means the next build could be the one, but I've gone on record saying that August of 2027 is when the scales tip and it is more likely than not that FSD (full autonomy, Level 5, remove the steering wheel, take a nap and wake-up at your destination) will emerge. That does not mean that it won't exist in limited forms (fair weather, geofenced...) before then, nor does it mean that it impossible before that date. It's just a probability distribution. If you want to know when it is 99% likely, move your bet to 2040.

Disclosure: I'm Long Tesla