Coyote Blog

You folks can disagree with me about the prospects for the Tesla stock price.  You may hero worship Elon Musk and think he's a genius where I think he is a charlatan with PT Barnum promotion skills.  You may really, really love our Tesla car (I thought the Model S was a fine car when it first came out -- the Model 3, not so much).  But PLEASE do not rely on your Tesla autopilot.  Or at least read this Ars Technica article first.

A while back I went to a party in the suburbs of New York and needed a ride back to my hotel in the city.  A friend of a friend offered to give me a ride.  It turned out he was driving a Tesla Model X (SUV).  In the same way that you can't go 3 minutes with a Vegan without getting a diatribe on the virtues of veganism, for some distance in the ride we heard a paean to Elon Musk and Tesla.  This guy had really drunk the Kool-Aid.  Before my wife started kicking my ankle, I expressed a few of the reservations I had about the company and Musk, and the driver got really defensive. So much so he set out to prove Tesla's superiority.  So he drove us the rest of the way home (mostly) with his hands off the wheel through the freeways around Manhattan.  It scared the cr*p out of me, because I knew that while the AP could drive impressive distances and navigate on its own, it was only 99% reliable.  And since it makes hundreds of decisions every trip, those are not great odds.

What worried me is that he insisted on using the AP almost as a test of religious faith.  There is this weird dichotomy where all the Tesla literature that is actually reviewed by their lawyers and the DOT tell you never to take your hands off the wheel, but Musk goes on 60 Minutes bragging and doing exactly that (almost hitting another car) and verbally the company constantly brags on its AP capability.

This is crazy.  You only have to look at Boeing and the 737MAX experience to understand that even a very careful and experienced company like Boeing can screw up the software-hardware interaction  The 737MAX worked most of the time, just like the Tesla AP, but in Boeing's case you don't see consumers arguing that this makes it OK -- you see the DOJ initiating criminal investigations.  Tesla is a much less experienced and far less careful and organized company than is Boeing.  Despite Musk's bluster, third parties rank them close to dead last in self-driving technology development, but they are the only major self-driving provider who are actively encouraging their customers to use it outside of a carefully controlled test environment.   Everyone who actually understands AI believes Musk is totally full of sh*t when he talks about AI, particularly the much-hyped "shadow mode."

Interestingly, that latter reaction is the same one I hear from nearly everyone who hears Musk discuss something they actually know about. My moment came listening to Musk talk about the hyperloop, which is truly a crazy, unfeasible, uneconomic joke.  As I wrote before:

Elon Musk is not the smartest guy in the world.  He is clearly a genius at marketing and brand building.  He has a creative mind -- I have said before he would have been fabulous at coming up with each issue's cover story for Popular Mechanics.  A mile-long freight blimp!  Trains that run in underground vacuum tubes!  A colony on Mars!  But he suffers, I think, from the same lack of self-awareness many people develop when they are expert or successful in one thing -- they assume they will automatically be equally as brilliant and successful in other things.  Musk creates fanciful ideas that are exciting and might work technically, but will never ever pencil out as profitable business (e.g. Boring company, Hyperloop).

I watched the Ant-man and the Wasp the other night and listening to Musk is a lot like Marvel movie physics -- both use recognizable terms (if you had a drinking game in this last Ant-man movie that took a shot when they said "quantum" you would be dead now) that sound good to laymen but make no sense to people who actually understand the topic.  There may have been some excuse to lionize Musk's brilliance a year or five years ago, but how can anyone think this guy is anything but a knucklehead with an overcharged ego after the Boring Company fiasco?

Anyway, don't rely on this guy's reputation.  If you like what you see in the Tesla showroom (if there are any left), then by all means by the car.  But do not turn on the auto-pilot.  Please.  A loss of even one of you readers could... reduce my blog visitation by whole number percentage points.

From a fascinating article on Amazon and its automation vision:

After a customer places an order, a robot carrying the desired item scoots over to a worker, who reads on a screen what item to pick and what cubby it’s located in, scans a bar code and places the item in a bright-yellow bin that travels by conveyor belt to a packing station. AI suggests an appropriate box size; a worker places the item in the box, which a robot tapes shut and, after applying a shipping label, sends on its way. Humans are needed mostly for grasping and placing, tasks that robots haven’t mastered yet.

Amazon’s robots signal a sea change in how the things we buy will be aggregated, stored and delivered. The company requires one minute of human labor to get a package onto a truck, but that number is headed to zero. Autonomous warehouses will merge with autonomous manufacturing and delivery to form a fully automated supply chain.

I got some cr*p on twitter a while back about writing this, but I think it is pretty much vindicated by the "one minute" factoid above:

Amazon likely is being pressured by the tightening labor market to raise wages anyway.  But its call for a general $15 minimum wage is strategically brilliant.  The largest employers of labor below $15 are Amazon's retail competitors.  If Amazon is successful in getting a $15 minimum wage passed, all retailers will see their costs rise but Amazon's competition will be hit much harder.

A few random notes on computer games for those who share that interest:

1. For those Diablo fans who loved Diablo II but were disappointed that Diablo III was not exactly the sequel they'd hoped for, I have a suggestion:  Path of Exile from Grinding Gear Games.  It is set up as an mmrpg (so you have to be online to play) but it plays just fine single player and all the map areas are dedicated instances such that you aren't fighting other players for kills and loot drops.  The skill tree is famously enormous.  A certain group of you will buy the game 2 minutes after clicking on the next link (I did).   Here is the whole tree, it is absurd (the highlighted areas are the selections for one of my characters).  The customization ability is simply staggering.   Choosing a class like fighter or mage (they have different names in this game, but essentially these base classes) just changes your starting point on this map.  But this is not the end of the customization.  There is also an elaborate skill gem system where your attack and defense skills are based on your gem choices, both the main gems and support gems one adds to it.  Seriously, the actual combat is not much more elaborate than the debuff then hack and slash and loot drop of other Diablo style games, but this game has more ability to fine tune and experiment with character design than any I have ever played.
2. My absolute favorite, by far, board game has finally come out as a PC game -- Twilight Struggle.  It is on Steam and I can't yet fully recommend it because I have not played through all the way online.  I am told the AI needs to be tougher but it should be fine for noobs.  There is also online person to person play.  I love the gameplay and it has also been a platform for my son and I to have a lot of discussions about recent history.  If you are a total noob, here are a few lessons for the Soviet player (which I have the most experience playing)
   • The Soviets have to rush.  The game has three periods, and you have big advantages in period 1 and disadvantages in period 3.  You HAVE to build up a lead early or you are toast later on.  I have seen a 15 point lead evaporate in the last third of the game.  The best outcome is to win the game outright by the smear rule (20 point lead) by turn 7.
   • Your first move is to coup Iran.  Asia is yours in the early game if you succeed.  The only alternative is to first turn coup in Italy, but that is a riskier strategy and can only be justified if your first turn hand is really tuned to that approach.
   • Coup every turn ASAP.  Coups are your most powerful weapon (other than events) and couping first thing every turn denies that ability to the US
   • The space race is for dumping your worst cards, not an end in and of itself (always exceptions, of course).  Twilight Struggle's best dynamic is how you end up with your opponents cards in your hand that you end up having to play for them-- the space race is one way to dump the worst of these cards (e.g. grain sales to the Soviets).  Since the cards you can play become more restricted as you advance in the space race track, there are even some advantages to failing your rolls early on.
   • If you play the China card, it needs to be for a BIG goal - like improving your scoring of Asia right before you play the Asia scoring card.  In many cases, it is better to not play the China card at all than to have it pass to the Allies.
   • Cards that allow you to play influence on any country should be used to get access to places where you have no adjoining influence -- don't use it to add to existing influence or enter countries to which you already are adjacent.   This is the only way in initially to places like South America and much of Africa..  Decolonization is your friend.
   • Learn to love this site.  Not only does it give you a LOT of strategy, but it also answers complex card interaction questions for every card.

Can we please make sure no one is able to put an AI into this thing.   We definitely don't want it to become self-aware.

Maybe I have watched too many movies, but it just does not seem like this will end well:

In its recently released "Unmanned Aircraft Systems Flight Plan 2009-2047" report, the US Air Force details a drone that could fly over a target and then make the decision whether or not to launch an attack, all without human intervention. The Air Force says that increasingly, humans will monitor situations, rather than be deciders or participants, and that "advances in AI will enable systems to make combat decisions and act within legal and policy constraints without necessarily requiring human input."

I will eschew some obvious Terminator clips and go a little old school

The original version of this scene actually did not make the theaters because it was too violent for the time.  If you want the full gore, .  The clip begins with the theatrical release, and then it replays the whole thing with the deleted bits.