His genuine interest in such a vast range of topics and his ability to keep experts away from rabbit holes made the show exceptional. He will be sorely missed
It’s impressive that Airbus caught up with Boeing after a 20 year head start. It sounds like Airbus’s bet on the future paid off but the article reads more like a PR piece than a case for why the A320 out competed the 737.
A320 and the 737 were designed in entirely different worlds.
The 737 was designed using light tables and slide rules, to use low-bypass turbofans and direct controls with avionics only on board to optionally aid the pilots.
The A320 was designed in CAD and using CFD, with full digital fly-by-wire, and designed from the start for high-bypass turbofans.
Both designs have been updated plenty since, but because the basic design is much more modern, the A320 is much more amenable to being updated. There are elements of the 737 design that still exist on every new MAX coming off the line that would completely doom the certification chances of any new design, but are still there because they got grandfathered in for 737.
The wonder is not that the A320 finally caught up in sales, it's that the 737 can still be legally sold.
> There are elements of the 737 design that still exist on every new MAX coming off the line that would completely doom the certification chances of any new design, but are still there because they got grandfathered in for 737.
Not only that, but Boeing is actually limited in how much they can "modernize" the 737, because doing too much might exceed the limits of the 737's type certificate. This is the reason behind the current engine inlet overheating worries, which has led to an airworthiness directive for the 737 MAX (https://aerospacenews.com/faa-airworthiness-directive_boeing...) and is also one of the reasons for the delay certifying the MAX 7 and MAX 10. This would be a complete non-issue for other planes, because all modern designs have a switch position that only turns on the engine anti-ice system when it's needed, but the 737 MAX can't have that because the 1967 737 didn't.
This sounds odd, they're able to certify the crazy MCAS but not a simple anti-ice switch?
I know that it's a complete nightmare to certify anything. However I apparently don't understand some underlying principle that allows to certify some things and not the others.
The constraint here (entirely self-imposed by Boeing's sales strategy) is that the newer models have to basically behave like the older models to avoid needing a new type certification. (See https://en.wikipedia.org/wiki/Supplemental_type_certificate) The aircraft behavior, and pilot procedures and training have to be substantially similar. That was the whole point of MCAS, to make the plane fly as-if it were an older model, despite the significantly different aerodynamics. Changing anti-icing procedures is apparently too big a difference to maintain the original type certification.
> That was the whole point of MCAS, to make the plane fly as-if it were an older model, despite the significantly different aerodynamics.
MCAS was implemented to make sure the control forces increase going into a stall, this is a requirement in the regulations. Without MCAS the control forces would drop on the way into a stall, which is an issue that would prevent certification of the aircraft.
From my understanding, mostly based on Kerbal Space Program, the aircraft isn’t well balanced when equiped with modern engines.
So you have to constantly apply some controls to fly, done by software.
I love stupid car comparisons so imagine a car with a new engine that is more economical to run, but very heavy on the left so the car constantly want to turn left. But if you apply force to the steering wheel manually or the car does it for you with software, all good. Still a shit car though.
The main issue arose because Boeing wanted to install larger, more fuel-efficient CFM LEAP-1B engines without changing the aircraft’s landing gear height too much (which would have required expensive redesigns of the fuselage and systems, triggering a new certification process). On earlier 737s, engines were already mounted quite far forward under the wing because of the aircraft’s low stance. The larger MAX engines could not fit in the same place without scraping the ground. Boeing moved the engines further forward and higher on the wing. This changed the center of thrust and lift characteristics. At high angles of attack (nose up), the repositioned engines created extra nose-up pitching moments, making the aircraft more prone to stall. To make the MAX “feel” like older 737s (so pilots wouldn’t require expensive retraining), Boeing added software — MCAS. MCAS automatically trims the horizontal stabilizer nose-down if it detects a high angle of attack, countering that engine-induced pitch-up. The tragedy was that MCAS initially relied on a single angle-of-attack sensor, so a faulty reading could (and did) trigger repeated nose-down inputs, leading to the two fatal crashes (Lion Air 610 and Ethiopian 302).
I think they added redundant sensors which should theoretically prevent this in future. IMHO, I think several issues compound here. They should have redesigned the fuselage. The engineering compromise is bad, but if handled with care, could have been done relatively safely. They opted for no additional pilot training re MCAS. This was a fatal mistake, compounded by them relying on a single sensor. Nothing in avionics relies on a single sensor for remaining in the air. That was insane. There MUST have been engineers screaming about safety who were ignored.
> MCAS automatically trims the horizontal stabilizer nose-down if it detects a high angle of attack
I remember when I first heard about the MCAS issue and thought "Why didn't the pilots just pull up?" and someone explained to me that MCAS worked by trimming, and it would trim nose-down so hard that even with full elevation to try to nose back up, it wasn't enough to overcome the extreme amount of nose-down from the trim.
And since pilots weren't trained on the new MCAS system, they weren't aware that the trim would have been automatically moved.
Boeing already had a angle of attack second sensor with the original design of the Max, but they ignored it and made decisions only using a single sensor.
How anyone thought making critical decisions on a single sensor reading made sense is beyond me.
Yeah, that was exactly my feeling too when I read that Airbus has "finally" caught up to Boeing. With that head start, catching up was not something that could have been expected (unless Boeing would have replaced the 737, which they arguably should have done years ago already, but that's a different story). Of course, if you look into the details, things get more complicated, since the 737 had an in-house narrowbody competitor with the 757 for some time - but Airbus now has the same, with the A220 competing with the smaller A320 family models (A318 and A319).
Was the 757 _really_ meaningfully a 737 competitor? I've only ever seen them used for fairly long-haul flights; didn't think there was much role crossover.
Boeing had a 20 year head start, but in a rapidly growing market that doesn’t count as much as in a stable one. FTA:
“By 1988, when Airbus launched its new A320 narrowbody jet, Boeing had already established a strong lead, having delivered around 1,500 of its popular jets”
So, Boeing sold about 1,500 767s in the first 20 years and over 10,500 in the < 40 years since. Airbus sold about 12,000 in the < 40 years since.
So, to catch up, Airbus produced about 15% more than Boeing did.
And will anyone talented enough to put all the pieces back together want to work for the federal government after all this? Feels like a good way of ensuring the only path forward is to pay large sums to private enterprise to fill the gaps.
That's assuming that the expectation is that the government keeps the same services afterwards. It seems the strategy is closer to making services dysfunctional to the point where they're unsalvageable, then cut funding for them because they don't work.
The schadenfreude is strong here but it’s hard to argue with results. By treating space launches as a process to refine through iteration (so the more the better) they’ve come to dominate the market. It’s also a process where others don’t gain the benefit of those iterations, there’s no new tech innovations to base your own launch vehicle on, just lots and lots of internal knowledge.
It’s worrying to have a capacity like this so concentrated in one company (whatever the leanings of the figurehead) but hard to see how that changes unless others start using the same methodology
> worrying to have a capacity like this so concentrated in one company
If it wasn't US spending their precious $$$$ on this (Elon says that in the end of they day these are ICBM's going to space), none of this would have happened.
I think that this deal "US gets super-duper ICBMs" and "Elon takes us to space" is a great deal.
I don't think any company's wallet can afford this, and Russia/China are decades away to do this on their own. Sharing technology is (so far) out of the questions because of the certain weapon-isation of the tech.
So for the next few decades it is Elon or nothing.
I think you're really discounting China. They have the resources, motivation, and the track record to fast track this. They've already landed rovers and have their own space station. They can fail faster than the US can succeed because of sheer size.
But ... Elon Musk has never once "come to dominate the market". He's always worked to figure out how to get government to give him ever larger amounts of money. He's built it up to the point where he gets close to 9 million USD per day from the US government (that's a stat only from his public companies. It doesn't include starlink military spending and the military part of SpaceX). Elon Musk's companies are something like Northrop Grumman/Lockheed Martin v2.0, not market-based enterprises, which they never were.
Tesla: cheating to make electric cars affordable, not by better engineering (in fact arguably way worse engineering), but by having the government pay (initially) 8000 out of a 45000 car, or close to 20%. Then continue by attempting, and failing spectacularly in an almost comical manner, to monopolize the battery market.
SpaceX: take a fundamentally bad business idea (there's plenty of private space launch attempts, but even the best go bankrupt), by ... getting the government to pay for space launches, in what is a very very bad deal (the government MUST have space launch capabilities. So it can never stop developing Boeing SLS. Therefore what is paid to SpaceX is paid ON TOP of what we pay to SLS, and is NOT a better alternative (as Musk screams), and so not any kind of saving. We don't have the numbers, because SpaceX is private, but all other companies trying it are 10x or 100x removed from profitability. Why would SpaceX, even if it has a 10x cost advantage, survive? And it is very hard to believe it's cost advantage is more than 2x or 3x if it exists at all)
Starlink: this has been tried, and tried and tried again. It just isn't profitable. SpaceX LEO satellites need to be replaced, 7000 satellites, every 5 years. That's 1400 satellites per year, or about 40 launches per year (you can only launch satellites together that go into the same orbit with very few exceptions), at a (subsidized by the government) cost of 70 million per launch. OPERATING (not building) SpaceX costs 2.1 billion dollars per year, not counting personnel, actual data transmission and uplinks, development of satellites and terminals. That's a minimum. The total market for internet in rural areas worldwide is ... about 300 million. The only real market for Starlink is military, and that means Europe, Russia, China, ... can never use Starlink. And even the US can't rely on it for several reasons (it's got a huge target painted over it's satellites and you can bet your firstborn Russia and China and Japan and ... have hundreds of experts working on destroying Starlink quickly. Hell, I bet the US has people working on that, even under Trump, just in case)
Explains a lot of why mr. Musk is so desperate to get a large amount of control over the government doesn't it?
Btw: does it really need to be stated that
1) every billionnaire is a financial engineer, finding holes in government tax policy, and if they don't admit this, they're lying.
2) mr. Musk's entire empire is utterly dependent on government spending. Totally. 100% (because all these companies would go bankrupt without government money, they wouldn't lose 10% of their income. They'd be gone). Therefore the idea that Musk wants to cut government spending is a bit ...
>SpaceX: take a fundamentally bad business idea (there's plenty of private space launch attempts, but even the best go bankrupt), by ... getting the government to pay for space launches, in what is a very very bad deal (the government MUST have space launch capabilities. So it can never stop developing Boeing SLS. Therefore what is paid to SpaceX is paid ON TOP of what we pay to SLS, and is NOT a better alternative (as Musk screams), and so not any kind of saving.
Sorry to interrupt your nonsense, but Biden's NASA administrator Bill Nelson quoted a member of the Joint Chiefs as telling him that SpaceX had saved the US government $40 billion for just launching military payloads. <https://www.fool.com/investing/2022/06/05/did-spacex-really-...>.
I explained why SpaceX is a bad deal for the US government. They have strategic reasons why they cannot be limited to SpaceX. Hence anything "saved" by a SpaceX launch is illusory: they can choose to pay for SLS ... or they can choose to pay for SLS AND for SpaceX.
Which number will be bigger? The price of SpaceX launches doesn't even matter. Or in economic lingo: SpaceX launches aren't that much cheaper to the marginal cost of SLS, which is what matters.
EVERY private space launch company beats SLS on price. Nothing special about SpaceX on that front. Who has the cheapest launch cost? This might be a surprise, but Russia does (about 50 million dollars). Yes Falcon can carry more and is more modern, but all rockets are different (and they do have big ones too), but you want cheapest? Russia is currently unbeatable, mostly because they do close to zero development at this point. Obviously, you can see the problem with using Russia too.
> They have strategic reasons why they cannot be limited to SpaceX. Hence anything "saved" by a SpaceX launch is illusory: they can choose to pay for SLS ... or they can choose to pay for SLS AND for SpaceX.
There are other options than SpaceX and SLS. And SLS is not designed or destined to launch the majority of US government payloads as an alternative to SLS.
SpaceX's primary market competitors are ULA and Blue Origin. The US govt can continue to fund competition even if the SLS program goes way.
Ok, I don't fully agree, but it's beside the point ... that makes all government money spent on SpaceX worthless (because it cannot replace money that has to be spent)
Think of it like buying a car ... after you've already signed the contract to buy a new car. The contract signing means that all money spent on cars afterwards is money wasted, because you will have to complete the contract regardless. Likewise, buying a rocket from SpaceX does not let the government avoid spending on SLS.
> that makes all government money spent on SpaceX worthless
I don't follow your logic. I think you're conflating two very different things, the US government launch market and the Artemis lunar program.
The US government previously paid ULA as the sole provider for many launch services. (Well, before that it was Boeing and Lockheed but they consolidated into ULA).
Since SpaceX became a competitive provider, the US government has had a choice on new contracts between ULA and SpaceX in competitive bids. SpaceX has won many contracts by being cheaper than ULA, and ULA has dropped their own prices to be more competitive to SpaceX.
That is real savings on US government launches through competition. There's no 20 year committed car loan on all future rocket launches that the US government is already paying for these services.
SLS isn't really part of the picture for the vast majority of these launches that SpaceX is winning contracts on. SLS is being built for upcoming lunar missions only at this point, as part of the Artemis program.
SpaceX's existing Artemis contract is to provide a lander to work alongside the SLS rocket and Orion, not as a replacement for SLS. Without SpaceX's lander there will be no way for astronauts to get from lunar orbit to the surface for the initial planned landing missions.
But treating SLS solely as money already committed to be spent is a sunk cost fallacy. The SLS continues to cost billions per launch even if you ignore all of the development costs up until this point. It can also only launch once every year or two.
A mission architecture based on competitive bids from multiple service providers (ULA, Blue Origin, and SpaceX) will almost certainly cost less going forward and allow an increased flight rate.
TBH if Musk is truely fixated with Mars then even 100 billion $$$ global commercial empire across multiple sectors wasn't going to finance it. The only play for trillions for Mars is to insinuate way into government largesse, and there's only one gov who can both afford to and willing (or dumb enough) to play. And even for US gov it's a squeeze... billion here, a billion there, and pretty soon you're talking real money... if Musk wants trillions no wonder he's wants to cut billions elsewhere.
The whole new iteration of attempts to game this are going to be fascinating. The adversarial nature of AIs trying to produce content better able to fool the AI judging the content sounds like the perfect Petri dish to quickly increase sophistication.
This is one of the many initiates where the process and details matter. What’s the due process here, can people see the prompt that judges them to be supporters? Is AI just highlighting “supporting” content for human review?
Even if you support the idea in principle, the actual practical implementation should be treated with skepticism.
Even if it is intended as a hint to human observers, during a big scandal in my country a few years ago turned out all the gov officials where blindly following the computer advise. Many where falsely accused of fraud by tax authorities, almost all where from ethnic minorities.
There's no "due process" for US visa applications. The process is, possibly deliberately, kafka-esque. Edit: I wrote about my visa process here: https://news.ycombinator.com/item?id=23942466
I’m not sure why you're being downvoted for this. the article does not state that the AI tool proposed to be used is an LLM. in all likelihood it is, but why assume?
Not that I think TikTok is particularly spotless but this sounds like the result of an algorithm learning what groups will engage with, ie democratic leaning accounts were more likely to do _something_ with republican content than vice versa.
Yeah. I remember Facebook used to appear to show me more of certain stuff that annoyed me (not political stuff so much as scammy stuff) because I hovered or even reported it. The most naive and high-level keyword based association is going to link both pro and anti trans stuff as the same general category too (another Facebook thing was the massive overlap between history and really dumb "alternative history")
This is interesting but I still think we have some responsibility here. What if most people hold some view in a low-grade kind of way but algorithms like this could accidentally (or intentionally) amplify it to the point of genocide. The story of Rwanda is well known and we can imagine it being much worse if we're not careful with this kind of technology.
You’ve been given the chance to show that your previous success wasn’t just a function of the domain you were in and team you were on.
Taking a flailing org and being visibly a part of turning them around will open a lot of doors in your current company. Notably those open doors won’t really translate if you switch jobs. If you switch jobs you’ll have to rebuild the trust that senior middle-management have in you.
At the end of the day if you want to find a small niche and stay in it then senior staff+ is likely not for you unless your technical area is in demand and very complex.
I'll also share a little of my brush with management... there's "easy mode manager" and "hard mode manager".
"Easy Mode" is when you're naturally promoted from IC to Manager in a problem domain you know, and already have the respect and admiration of your peers.
"Hard Mode" is when you're transitioned to manage a team where you don't know the problem domain, and don't already have a good working relationship with the people you'll be managing.
Much depend on your personality and support structure. If you're a "technical homebody" or don't have good support/rapport with the new director? This would be signals that this new role isn't the best fit for you.
The risk of deployments isn’t entirely technical. Depending on your business and customer base it might be necessary for some groups to have access to the feature earlier or later than others.
Whatever problems or limitations the existing approach had dropping everything on the floor is one of the least helpful ways of trying to fix it (assuming good intent).
Burning everything to the ground is a way of demolishing something though.
And if your intent is to just destroy it, it’s a far more effective one than bringing in experts to slowly try to disassemble the giant jenga tower without it falling over.
You have to assume competence too. You may have good intent but that doesn't help if you don't really know what you are doing or are blinded by ideology or some wayward belief.
