adamjcook

@adamjcook@mastodon.social

Engineer focused on #Robotics, #ControlSystems, #SystemsSafety, #Manufacturing and #Simulation. #ManufacturingOpen Contributor. #Purdue Engineering alum. Living in #Detroit. He/Him.
searchable

This profile is from a federated server and may be incomplete. Browse more on the original instance.

adamjcook, to random

Ok. So recently, had published an "Impact Report" which contained a slide presenting some "data" that their and products "enhance safety".

And one article and one Twitter thread caught my eye in scrutinizing these numbers.

https://www.forbes.com/sites/bradtempleton/2023/04/26/tesla-again-paints-a-very-misleading-story-with-their-crash-data/

https://twitter.com/NoahGoodall/status/1651323363099553793

While the analyses and arguments in this article and thread are not necessarily wrong, there are more fundamental issues here that need to be surfaced in my view... so let's take a look.

adamjcook,

is a divisive topic. Really divisive.

Two-sides constantly at war - generally over the stock price.

And it is also a topic that is surrounded by laypeople that are not competent in systems - so fundamental concepts are often immediately lost.

Totally expected.

Safety-critical systems are an oftentimes very niche and inherently complex topic.

For the pro-Tesla camp, "data", any data, is desperately sought to punch back against critics of Tesla's safety culture.

adamjcook,

Here's the rub.

Safety is not at all about numbers on a page!

This makes the article and Twitter thread I posted above, again not necessarily wrong, but entirely moot.

Why?

Because Tesla does not have a systems safety lifecycle in place for their and programs and products… therefore, the program and product are structurally unsafe.

Safety is about continuous maintaining a process of quantifying and handling failure - both seen and unseen failures.

adamjcook,

Aside: What is this about "unseen" failures?

Check out my previous thread here for Elk users: https://elk.zone/mastodon.social/@adamjcook/110162950844417313

And here for non-Elk users: https://mastodon.social/@adamjcook/110162950844417313

adamjcook,

Ok. Getting back to it.

Take the recent Boeing 737 MAX scandal for instance.

Between May 22, 2017 (its maiden flight) and October 29, 2018 (the first fatal incident), nearly everyone external to Boeing would have seen a perfect 737 MAX safety record.

The "numbers on the page" of fatalities linked to the 737 MAX would have been zero!

But the aircraft was never safe... because Boeing did not maintain an appropriate systems safety lifecycle for the aircraft.

The numbers were moot.

adamjcook,

It really is a broader problem with regulators globally.

The collects (unaudited, effectively voluntary) "data" from automakers on automated driving system incidents that they happen to know about...

testing "data" is viewed as a substitute for independent scrutiny of automaker validation processes...

Everyone is collecting data, but missing the big picture.

It is simply awaiting catastrophic, avoidable injury and death instead of proactively preventing it.

adamjcook,

Lastly, it should be noted that what has published here is not really "data"... it is more like conclusions which cannot be quantifiably scrutinized.

Broadly, unaudited safety numbers provided by manufacturers should obviously not be trusted.

Human lives are on the line and that should demand a much higher bar.

Boeing, after the first 737 MAX fatal incident, adamantly insisted that the aircraft was safe.

Then a second fatal incident occurred under nearly identical root causes...

adamjcook, to random
oldrawgabbit, to random
@oldrawgabbit@mastodon.world avatar

My pic. I think he's saying, if I cannot hug you I'll hug myself.

adamjcook,

@oldrawgabbit Aww! 🥰

adamjcook, to science

If you are into the intersection of , and (and everything in between), my recommendation is to follow @skrishna (host of PBS's FAR OUT).

I particularly enjoy Swapna's TikTok and I recommend checking that out too, if you are there: https://www.tiktok.com/@swapna_krishna

Swapna's work is insightful, inspiring and always engaging... featuring highly-digestible tidbits of information on topics that I seldom knew even existed.

Great stuff.

khalidabuhakmeh, to random
@khalidabuhakmeh@mastodon.social avatar

🌶️ Spicy take: The space would not be where it is today if it wasn't for , , and and other partners pushing the DX improvements they have over the years.

adamjcook,

@khalidabuhakmeh Not spicy in my book.

A shame that we could, at least for the foreseeable future, break free of the horrendous developer experience though.

CrackedWindscreen, to random
@CrackedWindscreen@mastodon.online avatar

Have to say, I am struggling not to be totally despondent with the car industry right now. Maybe feeling ill isn't helping but for some time I have felt pessimistic that the industry has its customers anywhere near its thoughts, in a good way.

The attitude to security and privacy. The design, both internal and external. The foisting of responsibility onto the customer when the systems are destined to fail. Subscriptions. Prices. Telling us what we want. And much more.

adamjcook,

@CrackedWindscreen Mr. Stock Market demands it now and it was only a matter of time before some hot new automotive startup discovered that automotive regulators (perhaps regulators in general) were always nothing but an empty suit.

Now everyone must dance to the same tune.

adamjcook,

@CrackedWindscreen Indeed. Highly dispiriting.

And I hope that you get well soon, by the way.

anss, to random

Block diagram and transfer funktion for Electric motor. Purple is speed output yellow is position output

image/jpeg

adamjcook,

@anss Ah.

Or as I like to call it..."control system modeling before Simulink". 😅

josem, to random

I cannot conceive any of this product's life cycle. Like, who goes to work in the morning to propose the idea of removing orange from it's natural packaging, to put it in plastic? And who buys this?

adamjcook,

@Danielsand @josem Oh. I did not consider that… 😓

My sincerest apologies then on my flippant comment.

CrackedWindscreen, to random
@CrackedWindscreen@mastodon.online avatar

This is an excellent thread outlining the issues with Tesla, driver assistance tech and the fact that the car industry is palming all responsibility onto the public whilst knowing full well we are being set up for failure.

Read it and understand what is happening as this covers the EU and UK, not just the US.

https://mastodon.social/@adamjcook/110275319409180523

adamjcook,

@CrackedWindscreen Note also the "made up" terminology that the jurors used when being interviewed after the trial...

"Self pilot"

"Auto assist"

Note also the illusionary concept of "taking control".

In a partial automated driving system, like , the human driver is always driving... responsible for the exact same control as if no automated driving system was equipped on the vehicle.

To these jurors, a division of control seems to exist... but it is an illusion.

adamjcook,

@kentindell @CrackedWindscreen Yup.

It is that quite clearly.

And a constellation of ever-expanding, safety-deficient influences that are left to run completely wild by regulators.

Honest to God.

I see absolutely wild concepts and terminology, constructed out of whole cloth by the Community on YouTube alone, each and every week.

But, hey, Tesla puts some warnings on the HMI sometimes... that should just about do it. /s

adamjcook, to random

Alright, so last week, a jury in California rejected the claims brought against 's product.

I am not a lawyer and the detailed analysis of legal issues is orthogonal to the obligations of experts in educating the public and in criticizing regulators.

The responses from the jurors in this case are interesting, though... and very much expected.

Let's briefly break this down a bit.

https://www.autoblog.com/2023/04/21/jurors-in-lawsuit-say-tesla-never-claimed-autopilot-to-be-a-self-pilot/

adamjcook, (edited )

Right off the bat, we are in big trouble.

It all comes down to something subconscious, something intangible - mental models that develop unequally and unquantifiably across an enormous pool of untrained, unsophisticated human drivers.

It is impractical to rigorously train human drivers to safely use automation like we do in commercial aircraft... so the mental models must be left to run wild and are effectively shaped by marketing, YouTube videos, word-of-mouth and so on.

adamjcook,

Complicating things further is also a very powerful illusion of control that exists in pretty much every human driver - the thinking that just because one is situated right in front of the vehicle controls that they can control all downstream outcomes even if the vehicle system they are operating is safety-deficient.

This is a powerful lie that we all tell ourselves.

And all of this, naturally, will exist in jury pools that likely (by dismissal) have no experience.

adamjcook,

The fact is...

This case, this "bellwether" case that is being celebrated by some, should have never made it to a jury.

It did make it to a jury because the has so completely dropped the ball on even marginally regulating the marketing and design of these extremely dangerous and inherently deceptive systems - systems that, by their very definition, are structurally unsafe.

And the NHTSA is completely fine with that.

What does the NHTSA care?

They are not getting any heat here.

adamjcook,

I suppose that is all to be said about that here.

The does not have the skill sets or the interest to understand or really appreciate what I wrote here, but those are the facts.

exploits that and, in time, so will all automakers to varying degrees.

tailorderg, to animals
adamjcook,

@tailorderg Cuteness received successfully. 🥰

gwynnion, to random
@gwynnion@mastodon.social avatar

Some people will never leave Twitter because it's familiar, they're an established journalist addicted to feeling important, or because they're right-wing chuds who want to see Musk and Truth Social Redux succeed.

Some will migrate to Bluesky for Twitter without Musk. Some will stick with Mastodon on principle or because it's good enough. And then there's some also rans. But I don't think Mastodon or Twitter are going anywhere anytime soon.

adamjcook,

@donnodubus @gwynnion This seems a bit harsh. Quote Toots are on the roadmap and development continues on Mastodon as quick as resources allow.

https://joinmastodon.org/roadmap

It’s an open source project and is hardly complete.

Seems like the major “killer feature” for is that it just seamlessly dumps everyone into a single instance and hand-waves all the federation complexity for sometime later.

adamjcook, (edited )

@donnodubus “Choosing the correct instance” or choosing an instance at all was the first issue encountered during the big migration phases. That was discussed, at length, for months.

believes in “eventual federation”, in a generous prediction of their motives, which I do not believe is actually practical. People will just largely stay on a single instance.

Messiness and finding consensus is a necessary path if any lessons from Twitter’s “downfall” are to be learned.

adamjcook,

@donnodubus All I am saying is that, fundamentally, breaks no new ground and it seems that its design intent was effectively a clone of Twitter - which is immediately frictionless to many just seeking a quick "Musk free" alternative.

Fine with me, but I predict Bluesky users are running straight into the same buzzsaw.

and the , while far from ideal today on multiple levels, are at least putting the work in to change an already proven-to-be disastrous status quo.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • megavids
  • rosin
  • thenastyranch
  • tester
  • DreamBathrooms
  • mdbf
  • magazineikmin
  • tacticalgear
  • Youngstown
  • ethstaker
  • osvaldo12
  • slotface
  • everett
  • kavyap
  • JUstTest
  • khanakhh
  • ngwrru68w68
  • Leos
  • modclub
  • cubers
  • cisconetworking
  • Durango
  • InstantRegret
  • GTA5RPClips
  • provamag3
  • normalnudes
  • anitta
  • lostlight
  • All magazines