Jeff Carpenter

Intelligent buildings design engineer, VP of international engineering operations, Live for family, Love to travel, RV'er since my 30's.

[Jeff Carpenter @jeffcarp]

@rajuvamsi007 Short of running other ADAS systems in a simulation of that exact scenario, there are no credible conclusions about how another ADAS-equipped vehicle would handle the same situation.

[Jeff Carpenter @jeffcarp]

@mike_poret Totally agree. That's exactly how I would plan to use this feature also.

[Jeff Carpenter @jeffcarp]

If you're making an automotive navigation app decision based on the politics of the name of a lake, you might need to touch grass.

[Jeff Carpenter @jeffcarp]

@SawyerMerritt @grok Using the current state & defensible pace of science and technology as your basis (not the opinions of online commentators or the views of executives invested in the industry's stock) how many years away do you think Level 5 vehicle autonomy is 1) to rollout & 2) to scale.

[Jeff Carpenter @jeffcarp]

@fletch33 @moregainzs I respect your opinion about what should be everyone’s priority when buying an EV, but there’s data on what actually matters to consumers today. In McKinsey’s 2026 survey, ADAS ranks only 10th as a premium-brand differentiator today. It’s projected to be #2 in 10 years.

[Jeff Carpenter @jeffcarp]

@niccruzpatane @scheiwiller Focusing on raw miles is naive. Focusing on L2 vs. L4 miles is also naive. If we're talking autonomous taxis, what really matters is how effectively the data from those miles is converted into real, demonstrated on-road L4 performance.

Anything short of that is a TED Talk.

[Jeff Carpenter @jeffcarp]

@fletch33 @moregainzs For most people, the decision for what vehicle to purchase doesn't even come down to comparing the ADAS features between cars. Given that, I doubt that "my friend has a cooler ADAS than me" is much of a factor.

[Jeff Carpenter @jeffcarp]

@niccruzpatane @the_og_jmosser Mileage is an input. It’s not a measure of intelligence.

Ultimately, what matters is the result the system achieves with that data.

Whether that result is demonstrated in a supervised or unsupervised setting is relevant. That’s what the Community Note properly addresses.

[Jeff Carpenter @jeffcarp]

@fletch33 @moregainzs FSD take rate among Tesla’s own on-the-road fleet isn’t even 25%.

It’s an awesome product, but clearly most Tesla customers don’t consider it essential.

So why is it bizarre that buyers might choose another brand?

[Jeff Carpenter @jeffcarp]

@bscholl @united Please. The U.S. airline industry sells a commodity product and provides marginally acceptable service. No U.S. airline will become a national treasure.

[Jeff Carpenter @jeffcarp]

@Ollie_XVX He's referring to excellent programs like the Waukee Schools APEX program.

[Jeff Carpenter @jeffcarp]

@kent_simi @HW4Checker @RivianTrackr Understood but rewriting code doesn’t discard all prior R&D gains. Tesla still had years of fleet miles and the model training that provides, testing infrastructure, camera integration and knowledge of edge cases. v12 was a new architecture built on years of prior development.

[Jeff Carpenter @jeffcarp]

@kent_simi @HW4Checker @RivianTrackr The first FSD Beta was released in October 2020.

I’m referring to Tesla’s full FSD development timeline, not just its current-generation product.

[Jeff Carpenter @jeffcarp]

@AmpedToGo @TravisKetchum @Rivian @SoonishEV A USB-IF certified, E-marked, Thunderbolt 4 / USB4 cable would be an interesting test, but really no reason to use anything more than a USB 2.0 cable. It doesn't seem like the higher speed provides any value.

[Jeff Carpenter @jeffcarp]

@AmpedToGo @TravisKetchum @Rivian @SoonishEV It might be that the problem connection is using a cable that has the TX/RX SuperSpeed cable pairs. EMI might be causing noise on those pairs. A USB 2.0 cable doesn't have those pairs, so the connection negotiates without them, to pairs less noise susceptible.

[Jeff Carpenter @jeffcarp]

@HW4Checker @RivianTrackr Are they being slow/conservative though? It took Tesla about 6 years to develop point to point autonomy which will be about the same length that it takes Rivian if it's widely available in 2027.

As the innovator, that was a hugely impressive feat for Tesla btw.

[Jeff Carpenter @jeffcarp]

@RivianTrackr @ChadMoran This entire discussion is very different based on whether the comparison is a Tesla vs Rivian or Rivian vs everyone else. If you're coming to Rivian from a legacy automaker, Rivian's pace looks like a breath of fresh air.

[Jeff Carpenter @jeffcarp]

@Legal_Eagle_52 @rajuvamsi007 @QuincyEdmundLee Tesla, Waymo and other automated-driving systems are making us safer. 💯

What gets absurd is the need to declare one company “the winner."

Tesla and Waymo aren’t playing the same game today. Multiple companies will succeed in autonomy using different approaches.

[Jeff Carpenter @jeffcarp]

@rantlab Setting fanboys aside, is anyone really arguing that lidar is magic?

Am I interpreting you correctly that you don't think that the lidar benefit justifies the compute cost?

Not sure I understand the dead end comment either, given the massive decrease in sensor cost over time.

[Jeff Carpenter @jeffcarp]

@Legal_Eagle_52 @rajuvamsi007 @QuincyEdmundLee You say "My point is Waymo can only operate in geofenced areas while Tesla can operate anywhere" but that's a comparison between supervised autonomy and unsupervised autonomy. That's not the same test.