The SpaceX Thread

[demultiplexer]

Some of this 'uncertainty' sounds somewhat... exaggerated. I've worked on fluid and stress models for heavy earth moving equipment (not the engineering side, but the technical simulation side... I'm not an engineer) quite a while ago (12 years or so) that was incredibly detailed and had little if any variance to real-world testing. Air flow over surfaces including turbulence models gets into incredibly dense math, true, but it's fairly well predictable as long as the model is correct.

Sure, there could be some fundamental errors in the CFD models that result in the FH shredding itself at MaxQ or something similar, but I wouldn't put it at 50/50.

So, none of this matters for why rockets are so hard to make. The real reason is quite simply: design tolerances. The most devastating kinds of stresses on a rocket are coincident bending and shear, and those are quite hard to 'solve' even for large transients without expending a lot of material on the envelope. All material at launch has growth factors of 100+, so you really want to minimize material use wherever possible, especially on the first stage.

Most of the peak loading on rockets is not analytic at all. It's stuff like acoustic loading during launch (due to the ground effect) and during max Q when shock formation propagates and often resonates in the structure. These are loads that can constructively or destructively interfere with each other. Combine that with the inherent stochastic nature of the frequencies of sound waves you'll get and it becomes really treacherous to 'properly' design for the maximum loading you can expect. The maximum loading can be some resonant mode that happens once every 100 launches, and that particular loading can suddenly increase stresses on some subsection of your vehicle twofold.

So you do eigenmode simulations, you thoroughly diffuse the most critical shock front formation sites, you modify the speed of sound through your structure to avoid significant acoustic energy transmission and so on. But that only goes so far. The tools we have nowadays are really not that much different from the Patran/Nastran days of the 90s, just an order of magnitude or two faster. CAD software can't just 'model a rocket'. You have some abstraction of your rocket as a rasterized model, you enter some boundary conditions and constraints and then... you let it run. Some engineer needs to adequately design the model, the loading and the boundary conditions to adequately simulate what goes on. Often, this is done on e.g. a few-degrees section of the symmetrical rocket and just assumed to generalize to the whole structure. Not because that is the best practice, but because simulations still aren't fast enough to simulate on nyquist-satisfying grid spacings.

The rest of the aerospace industry has it easy. Airplane manufacturers just design their absolute maximum loadings at 2x or 3x of 'normal' flight conditions, because airplanes can be way too heavy and still economical (and any crash will cost them orders). Rockets don't have this luxury, so... they blow up. A lot. Fact of life.

 

[demultiplexer]

So saw in an article that the President of ILS (commercial Proton-M operator) is throwing shade at SpaceX

ILS and Khrunichev have strategically developed a family of Proton variants that provide the necessary flexibility at an attractive price and you won’t have to settle for used hardware

LOLZ. Nobody has an answer for the gauntlet SpaceX has thrown down except Blue Origin and that is a someday kinda thing so nobody has an answer for SpaceX today.

Hah... that is comedic. I suppose these people also never settle for driving in their car more than once?

 

[demultiplexer]

Within the science community at large, space elevators aren't really taken that seriously as a short-term option. The concept of a space elevator is striking and photogenic, so it's often used as kind of an 'in' to the concept of non-rocket spaceflight options, but from a technical perspective space elevators are one of the most expensive and technologically challenging options.

Over the past 40ish? years (basically since the Planetary Society and other formalized academic circles started seriously considering it, fueled by I assume lots of Isaac Asimov-style hard science fiction) a lot more types of surface-to-orbital systems have been devised. Some options that are possible with today's technology are:

- Launch loops (also known as Lofstrom loops). These are active support structures, meaning they need no especially strong materials to support themselves, but they do need a constant input of power. Because of the very diffuse nature of the energy stored in a Lofstrom loop, as well as the rather low terminal velocity if it ever came down, it is considered by some metrics to be quite a lot safer and certainly much cheaper than a space elevator. The Lofstrom loop is also a great introduction into active support structures in general.
- Mass drivers. Giant electromagnetic cannons. Seriously, what could possibly go wrong? This is almost certainly what's going to be built on the moon and Mars for deep space travel. It works a bit less well when you have an atmosphere and not a lot of elevation, so not the best idea for Earth.
- Skyhooks. Especially the rotating kind. Combined with solar-powered ion thrusters and atmospheric grazing for fuel, these can pick up suborbital spacecraft at low speed (e.g. ballistic missiles) and use part of their kinetic energy as well as their rotation to put them into full orbit, then regenerate their momentum slowly to stay in orbit themselves. Bonus points for multi-stage skyhook implementations to go from lower to higher orbits.

All of these things are, in some form or another, possible without great expense (although generally considered still in the billions-of-dollars range) and within very reasonable timeframes. Like, this wouldn't make that much of a dent in the global space industry budget. And some of these concepts are even better than tower structures, as they can impart essentially arbitrary velocities to spacecraft at any orbital height, instead of being limited to orbiting only at GEO.

I'm seriously optimistic about the chances of some of these technologies seeing real implementations in the next decades, especially when super heavy lift vehicles like BFR make lifting ~1000 ton skyhooks (in parts) possible for even a large company, not just large countries with decades worth of space agency funding.

Should I plug Isaac Arthur's youtube channel at this point?

 

[demultiplexer]

Wow, I just got cancer from their use of units. I think they mean metric tons with 'mT', but to me it just reads like milliteslas.

Also, the arguments used by Jack Schmitt seem really unconvincing. BFR is going to be rated for life support, costs are similar (and impossible to compare at this point anyway, they'll both go over budget by at least double-digit percentages) and who the fuck knows how reliability and scaling is going to be in practice. I know that NASA sometimes has this image of an organization built on its most influential members' pet projects, and this enforces that stereotype.

 

[demultiplexer]

With any surface-to-orbit system, you don't start out with the hardest one. We're not going to build an space elevator or orbital ring. This all goes gradually. Technologies really only take off when developing and deploying them costs a diminutive percentage of the global market expense. Like, if spaceflight were a $1T business, you wouldn't expect anyone to build a $100B skyhook. You'll have to wait until the market expands to at least ten times that and fill the gap with whatever tech fits in between - which is often just scaling up existing tech.

The start is just cheaper rockets. There's fundamentally no reason why we can't get to $50/kg with hydrocarbon rockets, and maybe even lower with hydrogen. At that point, the market expands to daily launches and super rich people space tourism and you start spending hundreds of billions a year instead of single to tens of billions. Why $50/kg? Well, it takes about $25 worth of fuel to get 1kg up to orbital speeds and you'd expect a mature technology to tend towards majority consumables cost. Fuel is the only thing that is fundamentally consumable.

With BFR, we're going to need in-space refueling, and it seems like Musk is hell-bent on building infrastructure for that. We'll probably see the first human-scale (not lab-scale) space station in a decent timeframe. This is where you open up the market in the trillions of dollars range.

Then we're either off to a moon base or a cardioid (multi-stage) skyhook. A moon base is more resource intensive to build, but more stable/easier given current tech and opens up more of space to future markets, most notably asteroid mining and renewable energy generation. Skyhooks are cheaper and easier to build (at that point it can be built for a couple percent of the total global expenses on spaceflight), but especially the first one will have very limited lifting capability and will barely increase traffic by itself, it's more of a springboard to a bunch of skyhooks. A moon base would almost certainly just be mass driver-based and solar powered.

With at least the inner solar system well-accessible, there will be a massive market opportunity if you build a rotating habitat. Again, O'Neill cylinders are out of the question at first, but if you can build a skyhook you can build a ~1000 person rotating habitat. This will allow for permanent orbital habitation and running actual businesses from space.

With a moon base and skyhooks in place, you start to have serious raw resource flows. Space solar, He-3 as well as precious metals from asteroids are going to completely transform how we do manufacturing and energy generation. If by that point we've cracked fusion energy, this is the point where active support systems start becoming a big thing. Before this, there are better and easier methods of getting into space than a space elevator, even an actively supported one. And we're already thinking at least half a century into the future here, unless the space race picks up much more speed.

 

[demultiplexer]

Using hydrogen on mars is probably much more doable than on earth, because you can just make a balloon-shaped (or many balloon-shaped) pressure vessels. That gives a tank weight advantage of about 2:1 on cylindrical rigid. Along with the ease of manufacture compared to hydrocarbons, that makes a hydrolox rocket on mars easier than fossil-derived rocket fuels on earth.

 

[demultiplexer]

OK it's a meme but also true: most people who make big unforced changes in their life deep into adulthood are the weirdos. The people who don't just not fit into society, but haven't been able to fit in after years or decades of trying. Not all weirdos are bad per se, but they are all weird in a way. People who climb Annapurna or K2, sometimes people with families to care for, are a little bit insane.

With (early) space tourism, you're going to capture the intersection bit of the venn diagram between weirdos and rich people.

 

[demultiplexer]

It is interesting that they only way to have a profitable comm network LEO constellation is to basically first build a profitable launch company and launch your own satellites.

Both are questionable, right? SpaceX isn't really profitable as a commercial company because of its heavy reliance on a legacy market to begin with (with lots of 'at any cost' military and public payloads) and Starlink is kind of an inherently fucked business model with their finances really only working at very high subscription prices (and retention) compared to ground-based internet that is probably better.

I'm not saying either SX or SL aren't useful or profitable, just that there are some very specific circumstances that they need to be profitable, crucially circumstances that rely heavily on non-competitive circumstances.

 

[demultiplexer]

Has anyone else lost their enthusiasm for SpaceX? I want to be excited for what's going on, but I just can't anymore, much though I love rockets. I mean, I watched, but I was hardly as enthusiastic as I was during the early Falcon 9 days, or even the early BFS days. It was especially hard to see Musk holding the child he's illegally avoiding service for and keeping from his mother.

I wish the boards of SpaceX and Tesla had the guts the board of OpenAI had.

I'm not even that worried about Musk's involvement purely because he just doesn't seem to have anything substantive to do with the company anyway, and hasn't for a loooooong time. Like, SpaceX from an engineering perspective has been doing the right things - and has explicitly not been doing a lot of things Musk said - for a decade or so.

But the working culture and general techbro attitude towards everything is very grating. SpaceX has recently become kind of the anti-aerospace company. Normal rocket companies are actually really environmentally conscious (typically being staffed by a lot of EO people), but SpaceX just fucks up major and rare nature reserves without any mention of it caring. They source their materials from the open market, with a few elements being notoriously from dodgy sources. They have a terrible high-pressure work culture, which from a working hours perspective isn't unheard of (academia isn't great either) but this is a few steps beyond working at NASA or Boeing or ULA.

If your goal is to change the world, but in the process you ruin the world you live on, I don't know if you're really that altruistically trying to do good. Leading by example is the way to go.

Musk is undeniably an asshole (and that is probably going to keep getting worse) but his title of chief engineer is deserved.

He doesn't need to know the most about every single discipline required for SpaceX but he definitely knows enough about each of them to shape the vision and make tough calls. Sometimes he gets those calls wrong but usually not.

As an actual aerospace engineer, he gets a lot wrong or is obviously ignorant of the engineering process behind choosing something. He can bullshit his way through rocket dynamics 101 but that's basically it. This is fine by the way, I don't expect a CEO or somebody so far removed from the process to have any real knowledge on the matter. Every facet of rocket engineering is something you can dedicate your entire life to and still find new stuff, and in a lot of cases the design space is just really large and there is no obvious best way to do something - you just have to work through the details or try stuff.

The thing SpaceX needs to be lauded for, and really the innovation of all innovations here, is that they're actually doing stuff. NASA, JAXA, ESA, they've all had really good rockets that even rival SpaceX's cost and surpass their safety record - but they don't exist anymore. NASA's and ULA's attempts at a new launch platform has been stymied, for political, funding and actual engineering failure reasons. Meanwhile, SpaceX has been steadily progressing. Still very slow, especially compared to the 50s and 60s, but at least stuff is getting done.

 

[demultiplexer]

There have been other rumors that SpaceX has been burning capital faster than it has been coming in.

Just picking this one out - rumors? SpaceX has always been a cash-burning machine, and the more they launch, the more cash goes up in flames. That's Musk's entire play, right? He makes mad money with Tesla and puts it into SpaceX to eventually die on Mars.

 

[demultiplexer]

Yeah, there's no reason to expect that Falcon 9 production and operations aren't cash-flow positive. Similarly, Starlink is believed to be cash-flow positive as well. Cargo and Crew Dragon development and production are wound up, so future flights should be high margin.

The development programs for Starship and the Lunar HLS are of course pretty costly. That says nothing about the marginal cash-flow impact of additional flights of their existing hardware platforms.

Cashflow positive in the space business isn't saying anything, though. That's kind of what I'm getting at - it's fine that they can pay for the fuel, rocket and staff, but the R&D cost ten times as much as all of that combined. At least from what I heard in my old aerospace circles, SpaceX's pricing to customers is much more a matter of trying to gain market share and volume rather than trying to become profitable in an overall company sense.

IMO that is actually the way to go in space, I'm not making a negative value judgment on them.

 

[demultiplexer]

You're right, phrased like that it doesn't make sense. It does read like I'm pulling it apart as opex and capex, although I didn't intend it like that.

With my first comment, I'm saying that the company as a whole, all the parts it needs to function, are very much unprofitable, and that losses only increase as they grow. SpaceX isn't suddenly going to make a profit as they scale the launch business up, and they are explicitly not even trying to do that. All the income they get is just to fill the gap between whatever it costs to do what they want to do and Musk's income stream. I was under the impression that this was common knowledge among space nerds. None of the commercial space companies ever make money, it's always dependent on massive subsidies one way or another. Also, in the pantheon of space launch systems, SpaceX isn't doing anything special or super-efficient as a company, yet they offer lower prices than anybody else. The only way you do that is by... well, subsidizing it with your own money in this case.

With my second comment, I'm cautioning of the trap of being focused on stats of part of a business. Part of a business may be cashflow positive, but if it's reliant on other parts of the business which hemorrhage money, overall you're not going to get anywhere without a net inflow of loans, goodwill, etc..

 

[demultiplexer]

This seems relevant
View: https://youtu.be/1I3dKEriVl8?si=zSe_BIrUtzDazs4U
I'm sure some of the constraints / problems in that video can be solved by throwing money / space at it but I'm not sure all of them could be

The only viable way to 'solve' the computers-in-space problem* is to create a little bit of Earth in space, most importantly shielding and fluid heat transfer. Second most important thing is to have redundancy and (in-situ) serviceability, it is absolutely necessary to be able to put entirely new hardware in space every 2 years or so to deal with changing standards. All of that is very heavy and thus expensive, so it hasn't been economically or even physically viable.

Pretty important to note that this won't be possible with Starlink, as the individual satellites are way too small to house, power and dissipate the power of a redundant routing server. It's going to require a whole new kind of satellite. Not to mention the laser-based inter-satellite comms links that also - at least with current tech - has a massive power budget (a ~40dBm laser, roughly enough for its current bandwidth budget, would require ~250W input power - and you'd need multiple to do mesh routing). I'm pretty sure they're waiting for space-capable diode-pumped lasers to start trying that out.

*(the problem being: you can't put commodity computers in space, they need to be hardened and tested for long times and basically custom-made, which means they'll be many generations behind the current bleeding edge, which makes it unviable)

 

[demultiplexer]

Anything you're doing with 10k concurrent connections is going to need a fairly powerful machine. Any caching you're going to do worth doing in space is going to require tons of ultra-low latency bulk storage that thus needs to be something commodity to make any kind of sense.

Think of starlink satellites as cell towers, and have a look at the kind of magic an average cell tower needs to work.

 

[demultiplexer]

Why do you assert that it needs to be ultra-low-latency? It's competing with at least the latency of a round-trip from the satellite to a ground station and back up - i.e. 10s of milliseconds. That's an eternity for even the slowest flash memory. It's in the ball park of a rotating magnetic disk's read latency.

Additionally, it doesn't actually need to reduce latency for serving the cached content, most of which is not latency-sensitive. It just needs to spare bandwidth on links that are hypothetically congested. Those could either be cross-satellite or ground uplinks. I'll note that decreasing traffic on those links also means reduced average latencies for other traffic over them, which might be more sensitive.

OK, Network topology 101:

• It has the added handicap of the round trip. That's not something that makes the timing less strict, it only makes it more strict.
• The whole point of caching is that you take away all of the downstream latency by having a low-latency node closer to the endpoint. So you typically expect a caching server to have latency less than the upstream latency. Say a starlink satellite typically sits at 5ms, then you expect the caching server to serve you in less than 1-2ms. If it sits at 100ms, you'd expect 20-30ms to be acceptable. This has to do with the statistical nature of lookups. A median 2ms lookup is going to have extremes of e.g. 10 or even 100ms, and you want your cache hits to be economical in like 99 or 99.9% of cases.
• Internally, it's most efficient these days to have single-layer storage solutions. If you can get away with it, all that's in a caching server is a big bank of nonredundant SSDs connected directly to the CPU's PCIe lanes. Once you add higher-latency or more complex storage like NASes, you need to add additional layers of caching internally to deal with that, and that's just extra power and complexity
• Caching these days is something you do for expensive content. Caching servers aren't serving up tiny images like the google doodle, they serve up entire movies. Dynamic content, the things people most frequently access (e.g. social media feeds), isn't easily cached but it's also not bandwidth or storage sensitive. And stuff you want to cache is security handshakes, which are the major source of slowdown in straight internet traffic these days.

 

[demultiplexer]

Since computer topologies are probably more familiar with you:

In a CPU, there's typically 5 or 6 levels of 'caching':

• L1 or instruction cache
• L2 cache
• L3 cache
• System memory
• SSD cache

and then you hit the actual storage

Every successive cache level sits further away from the CPU and thus can afford to be slower.

Topologically, Starlink satellites sit really close to the user. There is literally a direct connection from the user to the satellite, thus this is the equivalent of an L2 cache. Something that has to be very low-latency to be of any use at all. Because the next layer of caching is... on the ground again, about the same distance away.

--------

Now, what do you actually cache in an internet context? Obviously, that depends on the network design. Ground-based, you can have a lot of hops between geographically not that distant places, so it pays to cache anything that sits beyond some kind of latency or bandwidth bottleneck. In the context of Starlink, there's no major latency bottleneck as everything is literally line of sight. But there are bandwidth bottlenecks - the up/downlink. So you'd expect bandwidth-limited stuff to get cached, aside from the really timing-critical stuff like DNS and certificates. And that's one of those relatively hard things to do.

It's conjecture as to what actually makes sense for starlink to do. I'm not saying the entire netflix CDN has to go into space. But I am saying that the kind of things that get cached, that make sense to put near the user, aren't easily feasible in a satellite.

 

[demultiplexer]

The rocket equation is going to fuck you if you try to do that

 

[demultiplexer]

Nitpick: Orbital velocity at a geostationary altitude is just above 3km/s, far from the 11km/s required for escape. There's no hope of easily send those decommissioned satellites in an interplanetary trajectory.

I was initially confused by this phrasing, but the absolute orbital velocity doesn't actually matter - the delta-V to escape matters. You don't need to impart another 8km/s to a GEO satellite to get it to exit.

From GEO you need about 2.3km/s to escape
From LEO you need about 3.2km/s to escape

Fun fact: Getting from the surface to GEO actually takes more delta-V than to escape.

 

[demultiplexer]

I think the cargo starship will come in around $100/ton to orbit. Pad it a few multiples once you add in life support and everything else a person needs, but just hauling my ass to orbit would be $10,000. I know SpaceX thinks that when the system is mature that amount will be on the lower end of between $1000-10,000, including life support. The additional energy to a Mars insertion isn't really that much. Hopefully by that time there will be an Aldrin cycler (maybe made out of an assembly of Starships socked together) that would have all the power generation, food production, light manufacturing, rooms, and entertainment needed for a comfortable trip. The same starship that's carrying 100+ people packed in tight for $5000 a seat would be able to dock with the cycler for the trip to Mars, and then you'd get back on the passenger ship for Mars EDL. Actual cost to ship a person to Mars at scale could be under $10k. You could literally take a few years or decades and "work from home" on Mars and earn enough money to live comfortably importing supplies from Earth at $100-300/kg for shipping (that's not far off from how much it costs to ship internationally with UPS) (not really, but it feels like it).

First of all, I think you mean $100/kg, but let's take $100/ton.

This is very far-future wishful thinking. Generally, if we go by my favorite source of information on this kind of stuff, namely scifi spacefaring forums, you can theoretically go down to about high thousands of dollars on chemical rockets. The reasoning is roughly like so:

LEO is the lowest you HAVE to go to get something into space and keep it there for a reasonable amount of time. All the rest you can do with non-rocket propulsion. The rough upper limit of rocket mass is thousands of tons, maybe low tens of thousands (10X starship). Starship is already in the order of magnitude of the largest thing you want to launch for dozens of reasons. A 2/3/4-stage rocket weighing 10 000 tons can put about 250 tons into LEO and is close to optimum efficiency in that regard. Rounding up, that's 10 000 tons of fuel and the cheapest you can go right now is CH4+LOX which has never been lower than about $200/MT (at nearly free methane prices and LOX at energy input prices). Barring massive changes in the economy and fueling economics, that's $2M in just fuel and oxidizer. Even if you design a perfect rocket (cigar rocket, i.e. an infinite-stage rocket that perfectly adheres to the integral rocket equation) you don't get lower than about ~5000 tons of fuel for 250 tons of payload. So fuel, no other costs at all, that's $2M/250~$8000/ton.

This assumes the rocket and operations are free, which they aren't. SpaceX is still a tens-of-billions-per-year-operation even if they build zero rockets. All that needs to be paid, and they don't launch that many tons into orbit. Just ops adds tens of thousands of dollars per ton, not even mentioning the rocket itself. So as long as we use chemical rockets anywhere in the chain, anything below like $100-150k/ton is basically impossible. And the absolute floor is at around $10k/ton in a world where only fuel is considered (the only truly non-reusable part)

----------

Before we get into cheap spaceflight, what is the true bottom of the market right now?

SpaceX has carried about 8000 tons to orbit and let's say invested about $40B to get there (we don't know for sure, but it's at least in the ballpark). That implies they have at least charged on average $50M/ton to survive so far. Without hard evidence it's hard to verify, but given the awarded contracts they surely must be cheaper by this point, maybe as low as $20M/ton on average.

The cheapest recent launches have been well below that, but that's... like, I get how you can sort of say you can launch a Falcon for $20M (implied <$1M/ton), but if all SpaceX did was launch $20M Falcon missions, they'd go bankrupt. That doesn't actually cover the costs. That's more like a marginal cost estimate, but they are clearly subsidizing this with specialty missions, military launches, etc.

But it's clear that marginal launch costs are going to inch below $1M/ton. If we assume no big space infrastructure projects are going to happen anytime soon, that's about the end of it. To do it cheaper requires more scale-up that doesn't have a market right now.

---------

How can we get cheaper? Well, space infrastructure! 10k tons to orbit sounds like a lot to you? What about 100k tons just in fuel? That's the kind of launch quota you need to even start about thinking to put a robot-operated moon base in place. Another 100k tons just in materials per year to service it. And that's, also, roughly the limit of chemical rockets. At that point, you're able to put so much stuff into space that cheaper launch methods start becoming viable. For instance, mountain-based mass drivers are totally doable for ~$10-100B and have a 'fuel' cost of about 30kWh/kg - even with expensive electricity that's maybe $10/kg in fuel and at 100k tons per year, that's something you can pay off with hundreds of dollars per ton in repayments - orders of magnitude cheaper than chemical rockets. You still need some additional launch tech to insert something into orbit and do transfers, but that's something rockets can do very cheaply if they're stationed in orbit and don't need to get out of the gravity well to begin with. Altogether, maybe $20k/ton is possible that way.

Want to get to $10/kg? One of the first problems is that you need more than $10/kg in energy to begin with. This requires some kind of significant breakthrough in energy technology, something like nuclear fusion (the scifi kind, not the real kind) or megastructure renewables (orbital solar, etc.). Then you also need to be able to construct something that efficiently puts you into orbit but doesn't cost much, which is impossible as even the amount of concrete and steel you need for a launch loop or skyhook or space elevator is humongous. This is 100+ years out, if we ever get there.

 

[demultiplexer]

That's only a problem if it's more expensive than current disposable launches, which it won't be. It will be cheaper to launch a mostly empty Starship than a Falcon 9, much less the ridiculously expensive ULA and foreign rockets. If nothing else SpaceX will switch entirely to Starship because they won't have to keep making second stages and engines and discarding them.

Starship will only be cheaper if they can produce and launch it at a cadence that makes it cheaper, though. It's not an intrinsically cheaper craft.

That's what people (like me) are so worried about in this space race to nowhere: you kind of run out of things to launch pretty quick if there's no wars to wage or planets to visit. Launch costs can be anything, a satellite just by itself is still a 20-year, 100-million dollar project.

 

[demultiplexer]

Unblurring Gaussian blur is possible given enough resolution, but it is in effect (because of how compression artifacts and pixelation work) a lossy process so your results will vary


View: https://www.reddit.com/r/mathematics/comments/t61uyu/is_gaussian_blur_an_invertible_transformation/

 

[demultiplexer]

I find it telling that the SpaceX thread on Ars did not move during a Starship test flight. Musk really broke something in the Space Enthusiast's community.

Political partisanship is everywhere, even in the comments of the Everyday Astronaut's stream, who got scorned for supposedly not wanting to say Musk's name out loud.

With a company and leader that is SO diametrically opposed to what spacefaring enthusiasts have been all about for the past 60 years.... what do you expect? We all grew up on Star Trek and Asimov, you know. It's hard to square that with SpaceX and Musk.

 

[demultiplexer]

Anyway, reading Reentry got me thinking, is there no jobs within SpaceX that has regular hour and repetitive task? Because all those people interviewed by Eric for the book are on the cutting edge. Not all jobs in SpaceX are like that right? There's got to be someone supervising production lines of Falcon 9 second stage that works in shift and rarely do overtime, right?

Oh of course, there are massive amounts of 'normal' jobs, but SpaceX (and pretty much all Musk ventures) are somewhat special in that they kind of hire overqualified, young, commitment-free people that largely are willing to put in overtime all the time anyway. So as soon as you're in an operations or engineering-focused area, you do feel like everybody is putting in way more than 40 hours of work. And to be clear: you're compensated really well for that.

Nothing in aerospace works like SpaceX. It's an extremely conservative industry, so it has extremely conservative hiring practices as well. Young people do get hired of course, but if you're going to Airbus as eng or ops, you have a 9 to 5 and paid overtime is rare. On the flipside, if you need to make parts for Boeing, of course you have to do unpaid overtime all the time, particularly when making 737 parts. The high pay, young genius hiring practices of SpaceX and Tesla are unique and remarkable, so it's obvious that a book is going to focus on that. But of course, just because of how large companies work, only a tiny proportion of people are going to have that kind of work schedule. Catering, janitors, production and QA are all just regular jobs.

(I have 2 best friends working at SpaceX/Tesla. Yes, I am a very awkward person for them to talk to about their job.)

 

[demultiplexer]

I just finished Liftoff this weekend and it was interesting remembering what Musk was like before he decided to become so politically active. The benefit of hindsight does show how much almost everyone brought in on the "saving the world with tesla and humanity with spacex" mirage.

The startup world doesn't care about your personal life and that was shown in the book and from talking to people I know that work at spacex. in the book there was an anecdote about luring someone in with promise of a big vacation or something like that knowing that the person would be too busy to ever take it.

I'm just glad I know I'm not the kind of employee they want. I don't believe in the Mars mission and I value my relationships outside of work too much.

Back in the day, Musk was genuinely saying all the right things all the time. I and almost literally everybody at the TU Delft believed him, he was extremely popular and uncontroversially so.

In 2009, I was just getting disinterested in aerospace after seeing how immensely conservative and slow-moving everything is, how much it's just politics and no technology. All the problems we're talking about right now - the SLS debacle, NASA losing relevance, even the Boeing/MD woes - that was the overwhelming state of the industry. You get into aerospace to push humanity forward, to go to outer space, whatever it is, to do something NEW and exciting. Yet our courses were Java programming, linear algebra and working with '70s technology. In the lunch atrium hung a blended wing body design by Fokker that is still, now, ahead of its time - and it hung there since the inauguration of the building 20 years before I started my study.

Yet SpaceX was the new place that was hiring, giving you a whole lot of money to do breakthrough stuff and do all the things that were deemed scifi or at least too ambitious in lectures. Musk was sprinkling it all with near-term scifi lore, but the real hard scifi stuff that was 100% based on scientifically plausible stuff. Same with Tesla by the way, he was saying all the right power electronics and battery things.

I still believe Musk's 2009-2014 phase talking points. That's the good shit. I'd like that Musk back, please.

I also genuinely believe some people are perfectly suited for the kind of high-pressure, high-performance workloads at Tesla and SpaceX. My friends are those kinds of people and they THRIVE. It's not for everybody, but likewise a 9-to-5 is just too dull for a lot of people. Some people function better at 68 hours a week workloads, including the stress. Engineering is already quite a toxic kind of workplace and it attracted that kind of person.

Those quotes about people being lured in with big vacations... yeah, that did happen, people got outrageous offers, but everybody knew (even very early on) that the culture was extremely competitive and you never had free time. It was kind of on you if you believed the fairytale.

 

[demultiplexer]

OK, what's the actual contention here? Whether or not there was a leak? Yeah, there was a leak, obviously, I don't know how that could be controversial at this point.

Whether that has anything to do with the earlier leak issues or what specific bit of metal leaked is kind of unknowable unless SpX volunteers that information, and even then we're talking about a design with thin safety margins designed to show capability rather than competency, so stuff is bound to break all over. In terms popular on this forum: we're at a low TRL here.

 

[demultiplexer]

Starship hasn't reached orbit because reaching orbit has not been their priority. Their priority has been 100% reuse, which means finding a second stage design that can survive re-entry at orbital speeds. They have achieved orbital speeds. Obviously they have a problem with V2, but if they really wanted to make a point they could revert to V1 and make orbit with that.

I get that the propaganda is really strong with spacex, but please don't like... open your mind so much that your brain falls out.

Starship's stated goals before they started anything wasn't to blow up half the time. IFT-8 was not supposed to explode the same way as IFT-7, etc. etc.. It's clearly failing way more than expected and designed. It's not remotely successful, not even by Russian cold war standards.

That doesn't mean it won't eventually become something much better, but that potential future is still completely uncertain. It might also, like so many rocket programs, fail indefinitely and go away in favor of some new program. This is not a big deal, the only reason any of this is becoming such a big thing is because Elon is involved.

Speaking of which:

What's not clear to me is how long the endless money spigot allows that kind of program.

I've said this before: Musk has to be pumping money into SpaceX pretty much continuously for them to be burning so much material and R&D. So the spigot runs out when Musk goes away or his liquidity becomes tight.

 

[demultiplexer]

SpaceX raised loads of external investor capital to fund Starlink. That's now cashflow positive. Their Falcon launches for not-Starlink are also prodfitable. It doesn't seem like any of their liquidity is coming from Musk.

Of course, that's true, kind of forgot the other external capital. That's going to be a lot safer than relying on Musk, but they'll still need that external capital to do all the R&D and launches to get Starship mature. It's a multi-year process that burns many billions.