The general term, I believe, is self-regulatory-organization. The theory, I guess, is that the government sets the laws and says very generally, "no fraud", and "you have to write your own rules and make sure they're good", but offers little technical guidance otherwise. I think this could work well if there were very heavy penalties for failures to self-police, but in practice the revolving door between government and industry incentivizes slap on the wrist style punishment. It's a hard problem and I don't think we're doing well over the last few decades.

While I do study relevant US laws as required in my job (requiring familiarity with how law operates in countries where we operate), I am not an American, but I do understand certain things with regards to the US constitution and relevant case laws.

The Congress originally meets only for a few months' time, usually less than 6 months. This is due to the reality of the time, where travel is slow and representatives only receive a comparable salary to most people. Thus, it is exactly empowered to delegate certain powers to the executive branch. As someone mentioned, the administrative law is a cornerstone law and yet it delegates many powers to the executive branch.

In fact, said law and many, many, many similar (federal) laws have been upheld constitutional in the Supreme Court. There are certain powers that only Congress can do, and cannot be delegated to the executive branch, but it is clearly laid out in the constitution what those are (notably spending). Now I said federal because in certain states, the legislature can only delegate in very narrow situations (usually only in cases where lives would be in danger or in the protection of properties and where a need of immediate response is demonstrated).

So I'm confused why are you saying that is unconstitutional, in fact American history shows a very different answer. If you think that should be not allowed, you're entitled to your own opinion. However unless I read it incorrectly, the constitution, even considering the various amendments, is unfortunately not aligned with your opinion.

Sure it is. Article I of the Constitution says that all legislative power shall be vested in Congress. That means anything that has the force of law--and all Federal regulations created by executive branch agencies under the current US regulatory regime have the force of law; you can be fined or jailed for violating them--has to be passed by Congress using the process described in Article I. So any Federal regulation that has not been passed by that process--i.e., every one of them--is unconstitutional.

The fact that current US jurisprudence disagrees with that statement just illustrates how far current US jurisprudence has diverged from what the Constitution actually says. The status of Federal regulations is by no means the only example: current US jurisprudence says that Congress can regulate farmers growing crops for their own personal use because of the Commerce Clause; and that a city government can use the eminent domain power to evict people from their homes and turn the property over to a private development corporation (that ends up never developing the land anyway), and that counts as a "public use" under the Fifth Amendment.

While I agree with this as a matter of personal opinion, it is not at all the actual fact in our current regulatory regime. Federal regulatory agencies create law all the time. Look at the Federal Register; every regulation in there has the force of law and was written by a regulatory agency.

> you would inevitably end up with some busybody who is not accountable to the people massively over stepping their power and creating tyranny.

Which is exactly the situation now.

When people vote for politicians who say

  "government is wasteful",
  "there's too much red tape"

Letting an external auditors understand all the small technical details and test them independently will basically halt any progress. It might make this specific change safer but for the long run will slow down innovation and development of better and safer products.

This has some similarity to software development- we test much better but prefer to move faster to achieve a better overall quality and be able to fix issues faster and better

Would somebody asserting that about _their_ _own_ _job_ even make it past the interview in the private sector?

Here's some more info if it helps: https://en.wikipedia.org/wiki/Normative_statement

Oh hey look, I'm signing yet another contract with an arbitration clause when signing up for a cell phone service. Must be my fault that all of them require the arbitration clause...

Efficiency - or the lack of it - is absolutely not the point.

I believe it is foremost a complexity issue, and the MCAS(?) failure is just one of the many things that could go wrong, that actually did. Boing probably have more issues with the plane that is waiting to surface given their culture ...

It's a management function. If someone comes to you and says "We want to rewrite everything so it runs on a Raspberry PI powered by a hamster wheel" you don't need to ask about the engineering spec of the hamster wheel.

The Max MCAS was only marginally more plausible in overview. Details were never going to rescue it.

The FDA very carefully reviews the submissions, requires validation of tests and safeguards so data can't easily be manipulated. But the system is built on trust. If a company wants to manipulate, fake or exclude negative data they can. They'll likely get caught, but not always.

I don't know much about the FAA, but as someone who works in healthcare, I know the FDA conducts regular audits of medical device manufacturers.

They roll in for a week, request access to everything and everywhere, then pick a handful of areas (randomly) to do a full deep-dive. Generally, if a company is cutting corners, discrepancies will exist in many areas and they'll quickly spot one or more of them. I assume the FAA operates similarly.

Hard to believe that it was so long ago. Man, I'm old :-) https://www.chicagotribune.com/news/ct-xpm-1999-11-03-991103...

If it is the point that the FAAs job should be easy because the only oversee Boeing and Airbus, well that point is just ignorant.

Textron (Cessna, Beechcraft, hawker), Piper, Carlson, Viking, ... There's dozens of aircraft manufacturers (I have intentionally not listed those who make hit planes, but I think they too fall under FAA checks).

I would not actually be surprised if the small airplane manufacturers puts substantially higher load on FAA than Airbus and Boeing do.

Certifying new commercial models is but a small portion of what they do. By number of employees and budget I would bet that ATC is actually their biggest responsibility. That's not to say that they should just rubber stamp all that...

If the FAA had paid attention they would have seen a company desperate to compete modifying numerous basic characteristics of an airplane to the point of making it aerodynamically unstable, using a flight control system as a bandaid to fix this.

Not to mention being such massively cheap assholes that they literally didn't install warning lights in the cockpit to tell the pilots when the sensor their flight computer would use to override control inputs, had failed.

The alternative requires raising costs and slowing progress and innovation, which is a nonstart in a competitive environment such as Boeing vs Airbus.

So why not reformulate it correctly. In my ideology, it is not possible to solve this problem.

If that ideology would be out of the way, it could actually be tested pretty good via unittest running a simulation. So once developed those tests would actually be pretty cheap to run.

Its just this ideologic blindspot that prevents good safety.

Well money and time.

You speak as if you think that if only the great billionaires accepted a little less money then we could have safety.

How many lives are you willing to lose for a multi year delay to accommodate the FDA recreating every required test that the pharmaceutical company did? How many lives are currently lost because the current system doesn’t work?

Your suggestion is akin to voter id’s as a requirement to prevent voter fraud. How many legitimate voters are going to be prevented from voting due to the new rules to stop how many prior confirmed cases of voter fraud?

This isn’t a lack of resources. This is society deciding that the resources are better used elsewhere.

* Actual decisions on resource allocation made by regulatory agencies, which have a close partnership and history of employment with the industry being regulated.

It's different to say "society decided" vs 'a specialized subset of society, with tangled incentives, decided.'

At the end of the day, it's a spectrum from (no oversight) to (full, independent validation).

Boeing didn't want MCAS highlighted as a change, Boeing didn't want the FAA to independently discover it, and Boeing got all these things. Either by action on its part or by design of what the FAA did and did not independently verify.

That's a strong indicator we should shift regulatory posture further towards (full, independent verification). And while it may be cost prohibitive to shift all the way there, that's doesn't mean we can't shift closer to it.

The problem with the FAA vs Boeing is not that the system is designed around the (correct!) assumption that the public's and FAA's interests, on one hand, and Boeing's interests on the other are aligned. This failure cost Boeing enormously!

What was not aligned were Boeing's interests and Boeing upper management interests. The pervasive failure of our society to force CxOs to align with the companies they run and with society at large is much bigger than just in aviation.

That some pilot was indicted, but not the management he was responsive to, is a glaring indicator of this failing.

Unfortunately, they did not work as hard as they should have to ensure that it would not malfunction. Or to make sure if it did pilot's were aware and had a chance to develop muscle memory for it.

All of that was Management's push.

Having not had a public debacle around them, how much effort would it take for you to personally certify that MCAS is the only novel system of its caliber on these aircraft?

The FAA should have pulled schematics directly from Boeing engineering during certification. And the FAA should have someone with enough technical expertise and experience look at them. And that person should have said "The submitted information by Boeing doesn't include full details on this subsystem."

Whether or not the Boeing test pilot highlighted the system for the FAA is a red herring. It's the FAA's job to find this, regardless of whether someone points them at it.

If the FAA doesn't have the technical staffing or expertise to do this, then that's the problem. Charging the test pilot is necessary, but not sufficient.

If the FAA infrequently performs this work (certification of a new aircraft), then flex in retired expertise! You can't tell me there aren't qualified, retired candidates (ex-industry or ex-FAA) who would have signed up for a year or two review. And all the better that they don't have career incentives!

Essentially, this is the FAA charging Boeing for not doing the FAA's job correctly.

Me? What makes you think I would have information on this?

If I had to hazard a random guess, I would think that, if you're talking about the MAX 8, there have been enough leaks and testimonies and whistleblowers that any comparable system would have probably been mentioned somewhere. I have no idea either way, I haven't read everything that's gotten out. But I don't see why a regulator couldn't use various means to figure stuff like this out with reasonably high confidence.

That process would include validating all aspects of the mechanical specifications and changes of parts (reused already OK parts from the same authorized suppliers would be a quick check-off), mechanical engineering, electrical engineering, computer software, and any changes for maintenance and end operators.

Describing it fully like that, I believe the only benefit to 'type certification' should be training for the end users, but major overhauls should require retraining and that should be caught.

The FAA was aware that this system existed.

So it's probably most accurate to say "the FAA was aware of the system existing, but incorrect on the details of that system."

What would this accomplish? Instead of a total system time average of 10 years, you now get an average testing time of 10 years to which you still need to add the queuing time, so you're even worse off than before.

(The queueing time won't be five years with two servers in parallel, and I can't do the exact approximations in my head, but it'll be at least two years. In other words, by testing in parallel you worsen the cycle time from 20 years to at least 22 years.)

This is a good general rule: by taking on more work in parallel, you'll make the turnaround time worse. This is why lean consultants go on about limiting work-in-progress.

Also a call to learn some basic queuing theory! It comes in handy often.

1: https://www.aviationtoday.com/2021/06/03/faa-asks-budget-inc....

When there's a defined process that appears to run in isolation, I don't see why there should be only one queue in this case, considering that the task length cannot be easily reduced.

What I was saying with my previous comments was that giving the current budget levels, it doesn't help to shuffle people around (without some strong assumptions on the process, which in my experience rarely are true in practise.)

Most mechanical processes aren't necessarily conducive to parallelization. Verification and information processing on the other hand can do favorably in the presence of non-reliance on a physical system-under-test.

Which subsystem vetting arguably is. If you're talking vetting specs.

I'm assuming the FAA does this already. Human organisations have a bias to parallelise to a fault. Increasing the parallelism level beyond this does not improve lead times.

In that specific case probably the important parts to review are the bits that try to make the new airframe handle the same as the old one.

No, it isn't. It's just more work than the government feels like doing, involving more technical skill than the government feels like hiring.

The problem is that the government can't have it both ways. It can't both claim that it is regulating airlines and airplane manufacturers to protect public safety, and also claim that it can't independently check what the regulated entities are telling it. It has to be one or the other: either we get the actual independent regulation that the government claims to be doing, with whatever resources it takes, or we all admit that we are not going to get that because the government is incapable of doing it, and we figure out some other way of ensuring safety.

And aerospace engineers are dirt cheap. What sort of clown show is the FAA running?

That sounds like they are knowingly underfunded, which they are.

> And aerospace engineers are dirt cheap.

They are? Where are all the airline startups? I assume they measure in the hundreds with all this dirt cheap talent running around.

Aerospace requires lots of capital expense and is very ‘large customer’ driven and aerospace engineers play second (or third) fiddle to that, unlike in software.

I suspect you're letting your software world experience dictate your expectations of what engineering is all about, and what it takes to actually get work done.

Software development is a rare field where the only relevant resource is man hours. In other fields, including aerospace engineering, trained meat bags tend to have a negligible cost to the point where replacing a whole engineering team might be a minor inconvenience. However, crashing a prototype is a project killer due to cost alone.

Also, you're not hiring from a diversely employed Valley pool: most of the engineers with the requisite level of understanding to test Boeing's hardware work for Boeing (and to an extent its supply chain), which might mean you don't have to offer them much of a pay rise, but it also means [i] you're weakening the engineering capability of the firms actually designing and building the stuff by poaching them and [ii] their views on what's safe and what's an appropriate level of testing aren't fully independent anyway.

1) Boeing lays off thousands of aerospace engineers regularly (they did as part of the max disaster), and doesn’t rehire them all back - the industry is highly cyclical, and Boeing is periodically shifting locations anyway

2) the stated concern was Boeing brass was applying undue influence to engineering correct? If those folks worked for the FAA directly after being laid off , wouldn’t they be more than happy to stick it to Boeing brass if they were telling them to cut corners?

3) we’re talking design overview and identifying where Boeing (or others) may be ‘putting their finger on the scale’ or trying to snow regulators by asserting bogus test results or designing tests that they can pass by not including important test criteria they may not pass right? That is certainly something an engineer who was previously in the industry would be aware of, or even a independent engineer should be capable of spotting from ‘the outside’ - and require they do.

4) at (linked in a parallel thread) a median salary of $118k, which is well within something the feds could cover, the FAA can certainly afford to hire a non-token amount of aerospace engineers onto their staff if they actually wanted too/Congress wasn’t trying to kill them. This isn’t like hiring on a FAANG staff software eng for 700k or whatever which would cause outrage or break the pay scale, and this is for something for which there are clear large body counts that can be pointed at.

Now if we want to say Congress has been strangling the FAA for a long time (like the IRS and USPS) and forcing them to outsource to industry or whatever, hey - I could believe it - but that is something that should be yelled from the rooftops because that can be fixed, and that will cost us a lot in blood.

I don’t want more Americans dead due to corruption of a regulatory process, especially not my friends or family, and those are the stakes here.

It should, as it was what I said. I'm really not sure if it's possible to make a point any clearer.

> Boeing lays off thousands of aerospace engineers regularly (...)

Sounds in line with the classical big corp style of management. I'm not sure what any of that has to do with humans not being the critical element of providing a service. In fact, are you sure you're not supporting the point you're trying to refute?

And that Boeing would surely be exerting influence on them so they wouldn’t catch issues - after Boeing laid them off?

And that it wouldn’t matter having competent engineers at the FAA because catching things require expensive tests - that the engineers if they existed at the FAA could mandate Boeing pay for, since they would know they needed them to do them?

Huh?

No one is saying the FAA should be running a full parallel aerospace program. I’m saying if they lack in house competency to call bullshit on what a player they are regulating is passing to them and relying on that player to just always do the right thing, then they are not effective regulators

It would be like taking Facebooks word that they are totally being good privacy wise, and not having anyone available who understands internet tracking or adtech. Which, is of course another failing regulator (looking at you FTC), but at least that doesn’t get hundreds of people killed in giant fireballs?

Like almost every other agency (the FDA didn’t do the vaccine trials; are they incompetent too?) they don’t do the tests. They just police the industry.

Theoretically if you set laws and regulations and dole out severe punishment for bad behaviour, you don’t need to be the one running the tests.

People should want their government to run on trust (if their culture is compatible with trust) because it’s far cheaper and more efficient.

When will Boeing CEO and Board of Directors get locked up as part of the severe punishment?

This is naivette.

See this:

https://www.corporatecrimereporter.com/news/200/lead-boeing-...

From the article:

the case was settled with a deferred prosecution agreement — an agreement that Columbia Law Professor John Coffee at the time called — “one of the worst deferred prosecution agreements I have seen.”

Boeing did not have to plead guilty to any of the allegations.

No Boeing executive was charged.

And the Boeing deferred prosecution agreement included an unusual provision finding that a compliance monitor was not necessary because “the misconduct was neither pervasive across the organization, nor undertaken by a large number of employees, nor facilitated by senior mismanagement.”

“That is without precedent,” Coffee told Corporate Crime Reporter earlier this year. “I have not seen that anywhere else and I’ve looked at a number of deferred prosecution agreements. Prosecutors themselves are not conducting the investigation.”

Boeing’s lead corporate criminal defense law firm is Kirkland & Ellis.

Erin Nealy Cox, the lead prosecutor in the Boeing case, left the Justice Department earlier this year.

And last month she joined Kirkland & Ellis as a partner in its Dallas office.

Last I remember this being a topic of discussion, the stance was ‘trust but verify’ no?

While it may be efficient to rubber stamp things, it is not doing their job. Folks scam all the time, especially if they know no one is looking.

If they lack the competence to be able to independently verify, they aren’t being effective regulators.

Bingo. If I use 3 managers and PMs who don't know how to program for a code review, then that's not a very effective code review.

Inaccurate, please stop sharing misinformation. Seeing this on HN is unfortunate.

Most aerospace engineers are lucky to break 75k/yr to start with little to no equity, and often need to move to the middle of nowhere (compared to say NYC, SF, LA, etc for software), and get hit with periodic catastrophic layoffs with the regular cycles in the industry.

It’s pretty common that software folks are paid 2-5x with far less intense or zero credentialing and better work conditions - at the same company.

Median numbers, so masks variability. But it looks like AEs are generally paid decently, relative to other engineers.

https://www.bls.gov/ooh/architecture-and-engineering/mobile/...

That’s roughly half of the initial comp for an entry level software engineer at any of the SV firms, and most folks will be making much more than that at said SV firms within a couple years.

Being able to get a team of 4-5 experienced and credentialed aerospace engineers for the comp of a single ‘senior’ (mid-level somewhat competent but not amazing) software engineer sounds dirt cheap to me?

If you want to baseline off SV salaries, you shouldn't be looking at median all-AE numbers.

We’re on a SV startup website, where the comparison to cheap or not is of course going to be based on this.

You provided as a counterpoint to my statement on aerospace eng’s being cheap, data which shows median salary across all experience levels of the field being half the starting pay of a typical entry level SV software engineer - which typically requires no specific credentials, unlike Aerospace engineering.

If there is a large cluster of companies who pay 4x the median aerospace engineer salary to Noobs, then please provide said data. My understanding is those don’t exist.

SpaceX, a high profile name and maybe the closest to a SV type place you’ll get in the industry pays between $70-100k to their Aerospace engineers, based on multiple sites. Here happens to be a random Reddit thread about it in the first couple results.

[https://www.google.com/amp/s/amp.reddit.com/r/engineering/co...]

Which is exactly the point I’m making. When a straight out of school software engineer has a whole section of an industry they can go to that will pay them 2-5x what an experienced aerospace engineer mid-point or even late in their career can make ANYWHERE (except MAYBE a one-off consulting gig somewhere), then aerospace engineers are cheap no?

EDIT: Some required activities by the organizations to be certified can be outsourced to third parties, the org itself is still accountable.

That's an easy one. It's too big to fail. One way or another Boeing had to make it out the other end of the 737 Max disaster intact as an organisation. Anything else would have been unpalatable from the point of view of the American military industrial complex. I know that phrase is usually applied in a derogatory way but here I don't even disagree with the thinking.

Or they could be several times more efficient. $17.5 billion/yr can do a lot.