Right; SamuelA brought this up earlier. Ultimately, I think we’ll just have to accept that machine learning systems cannot be validated in the same way as ordinary ones. We’ll never have perfect introspection into their behavior. But they’ll be enough of an improvement over current systems that we won’t really care. After all, humans cannot be validated, and we let them drive. I’m sure there will be some regulation as to what constitutes an acceptable level of qualification, but ultimately lawsuits (and hence insurance) will determine the fate of these systems.
Of course the AI chips themselves can be validated–ASIC validation is top-notch, given that they have to run with an error rate vastly below any mechanical device. But the neural nets that the chips execute are a different story.
It’s possible we’ll reach a point where most cars are running the same (or almost the same) net. Apparently, a huge number of nets out there are just modified versions of one that Google trained a while back. The pre-trained version has much of the basic image recognition “infrastructure” already baked into it, and it’s a few short steps from that to a more specific version that just detects hot dogs or whatever. So we may end up with a baseline “driver” net that everyone uses because it’s the best known, and individual automakers just implement a slightly modified version on their hardware. This also gives some resistance to lawsuits since, in case of failure, they can just point out that they’re using industry-wide best practices.
Incidentally, building an “AI chip” is not particularly hard. Way easier than a CPU or GPU. Which means I think Tesla will be successful in their efforts. They’ll also have competition, though it may not matter since it seems the primary goal (as with so many of their projects) is to avoid getting dicked around by suppliers. EVs have lower requirements than, say, cell phone chips, in terms of power efficiency and other metrics. Even a moderately good effort could pay off a lot for them.
I don’t think anyone disputed that they’d eventually ramp up production rate. I’ve merely bristled at the idea of shipping anything to customers built in the way they’ve been doing it.
So far, 381 unique VINs have been spotted, but of course the real number is much larger. 251 were spotted up to Dec 1. Since it’s now at 381, that implies a pretty significant increase in output in the past couple of weeks. Particularly since actually spotting them is probabilistic, and the odds of seeing one is much greater for older production.
Of course there are a bunch of contributing factors that are impossible to estimate, so going from this to absolute numbers is basically impossible. But it does seem like there’s been a ramp-up.
The highest VIN spotted so far is >2000, though there’s no guarantee that there’s anywhere close to a tight packing. It’s probably a reasonable upper bound since higher numbers would have been spotted already if it were much higher (say 2500). It doesn’t say much about the lower bound, though.
Incidentally, this VIN collecting is similar to the German Tank Problem used in WW2.
Probably in that ballpark. There’s a large batch of Model 3s at the Fremont delivery center (according to comments, around 200). If that’s (complete speculation) half the past week’s output, that would be 400/wk or 1600/mo.
About the car itself, they had two of them at the L.A. auto show and I must say that in person, it’s a much better looking car than the Model S and X. The front and rear fenders are especially good looking when you seem them in three dimensions. As far as anticipation goes, it definitely had the largest crowds around it of any car there.
I like everything but the front end. I would prefer “something” in place of the non-grill. It looks like a clay sculpture that isn’t finished. I understand the need to prove it’s not an ICE car but it it is a let down from the rest of the car which has great lines.
The Model 3 pix I was able to Google up show a small grill down low in the air dam area. If I was them I’d have done anything to eliminate that.
Ref the P-40, the first airplanes without propellers looked pretty weird. Now ones with props look weird. Pretty soon cars with big grilles on the front will look quaint like this when we expect these machines to look like this.
The way to be the future is to fully embrace that future. If Tesla wants to be a leader, which they most certainly do, they should make a bold statement that they are a leader. They did it in many ways, but IMO they played a safety with the Model 3’s nose.
it still needs some airflow for the battery and motor/electronics coolant. You can’t tweet away the laws of physics. but even ICE vehicle grilles are largely cosmetic; they’re considered part of the car/brand’s identity. put aside the “unique” styling, but consider how little open area the Lincoln MKT’s grille actually has.
my only issue with the Model 3’s styling is that the front is obviously shaped to comply with Europe’s pedestrian safety regulations, and w/o a grille (fake or otherwise) to “hide” the mandated bull-nose it kind of looks like an old shoe.
Any car needs air intakes for some stuff; they’re not only for feeding an ICE’s coolant radiator. My point was that Tesla *could *have hidden those other intakes to make the ostentatious styling claim “Our cars are different; not primitive mouth-breathers like all those warmed-over 20th Century designs everybody else is still peddling.”
They chose not to do that and it seems to me, and I’m no stylist, that they missed an opportunity there.
Ref the MKT: Wow, that’s quite a set of baleen you have there Mrs Lincoln! My personal “favorite” example of this messaging is the Lexus; designed to appeal to Galactic-level Pricks. Once seen it can’t be unseen.
I knew Nissan wasn’t the Dark Side!(ref: that stupid commercial where the girl is daydreaming instead of driving…) For my money you can’t go wrong with NACA ducts and functional scoops https://i.pinimg.com/736x/2f/4e/86/2f4e86cae33fd31c4731b3b8e87a511e.jpg … There are lots of beautiful cars out there that dispensed with the open front nose like the Lotus Esprit or most Porsche 911s , but there are also plenty that look just plain ugly (I’m looking at you 1st Gen Ford Probe). Style is pretty much always going to be polarizing, especially if you’re setting up to challenge the status quo. Personally, I kind of like the Model 3’s looks (or the Mazda 3 for that matter) but I don’t like the upturned snout so much and it just looks like they plastered over where the grille was. Something more along the lines of a Maserati Mistral or a Lamborghini Muira would be nice… Not exactly challenging the status quo there though, or meeting pedestrian safety requirements.
Mostly only at rest, though–and more specifically, while Supercharging. The Model S has louvers that open when the battery gets too hot. No need for them to be in the front, really, except that they may as well put them there to make the car look more “normal”. It gains nothing from aerodynamic forced convection since the car isn’t moving when it’s needed most.
Regarding the model 3 : Found a couple of interesting charts here in this article :
First, how do you maximize battery lifespan? You do this.
Set the GUI in the EV to just 75% max charge unless you know you are going to be driving more than half the car’s range that day. See how the “charge to no more than 75%” beats every other case?
Second, if you buy an EV, be aware that if you don’t drive the heck out of it, you’re burning money. The batteryis degrading whether you drive it or not (table 3). You will not get 20 years, nowhere close.
Anyways, these are interesting tidbits. No, Tesla’s magical technology will not change these fundamental numbers, just push them up. Maybe they get a little more cycle life than the averages here since their batteries are higher quality, but charging above 75% still is going to do more damage. Maybe they have a bit better long term longevity (doubt it), but the battery degrades from time whether you drive it or not. (this is not true for ICE engines, at least not to any real extent, if stored properly, engines can last for many decades and need minimal if any maintenance to be started again. Other than needing new batteries and fluids, of course)
every hybrid and PHEV from the original Prius has done this. IIRC the gen 1 and 2 Prius keep the battery between 30 and 60% charge. the gen 1 Volt had a 16 kWh battery and only let you use 10 kWh.
that way as the battery ages and naturally starts to lose capacity, there’s no loss of range.
Looking at table 3, the one captioned: “Elevated temperature hastens permanent capacity loss. Not all Li-ion systems behave the same.” and I’m not seeing what you’re seeing. Oh the idea that capacity degrades with time driven or not sure. That’s pretty well established. But anything that allows for any prediction of how long these specific EV batteries well oversized for daily commuting will “last”? Anything that substantiates a claim that owners won’t “get 20 years, nowhere close”? Not seeing it.
Reality is that we have no way to know whether or not that trend line continues tat flat or falls off a cliff at a certain age. Either could be I guess. But then again one also has to define what capacity one is willing to accept. Even if that trend line steepens with time … starting off with 220 (let alone the 310) range can allow for significantly reduced capacity to still be adequate for many users needs.
I see nothing in the real world data that makes me doubt that 20 years plus of real world use is unlikely for most users.
