4dpuzzle

I know I’m going to be bludgeoned for this. But I’m mystified by this ‘iterative’ approach to development. I wonder if they got this from the IT industry. Charles Bolden once remarked that the congress would have shut down the Apollo program if they lost vehicles at the rate spacex does now. These failures also often feel like the result of lack of foresight and critical thinking, rather than the consequence of complex chains of events that are hard to predict.

Consider the first flight. They decided to launch it without a flame deflector or a deluge system. They thought that it would be OK based on a hot test of the superheavy at half thrust. I don’t think any other rocket company would have made the same decision. Even if the concrete slabs didn’t shatter into a thousand pieces, the reflected shock wave would have been damaging enough to the engine compartment. They predictably lifted off with several failed and failing engines.

Another problem was the stage separation maneuver. They had planned on a full 360 degree cart wheel to separate the upper stage with centrifugal force. Not only was it going to cause enormous lateral and bending loads on the two stages, I’m still confused about how they would separate at all when the airflow is pushing the top stage (whichever happens to be on the top) against the bottom one. Perhaps the sideways loads might snap the joint. But the two stages are still in danger of collision due to airflow. This concern was proven to be valid when that flight cartwheeled several times without separating and then buckled. I don’t know if that’s the reason why they abandoned the maneuver, but something was definitely wrong with it. And it was bad enough to switch to hot staging.

I don’t have many comments about the second flight. Both stages failed in flight. The first stage failed due to problems with filters. I’m willing to give them a pass here.

As for the third flight, the superheavy exploded 400m above the seas. It was clear that it didn’t decelerate enough due to engine failures. More engine problems. But a pass here too, since engines are generally hard and the raptor is particularly nasty at that.

The real problem in that flight was the starship. It was tumbling pretty badly even in space. The video didn’t give any clue about how they could arrest the tumbling. I was looking for the operation of attitude/RCS thrusters and I couldn’t find any solid evidence of any of it. It was more like the starship had no attitude control at all, rather than one failing. Perhaps I’m completely wrong and it had attitude thrusters. But it was clearly deficient at least. The commentators of flight 4 said that they added more thrusters for attitude control. That would mean that either they didn’t consider tumbling as a problem or they completely underestimated it for flight 3. Why? Attitude control isn’t the hardest problem in rocketry. Very good simulation and analysis techniques exist for it. Anyway, a sideways reentry is bad enough. Even worse is an unarrested attitude rate at reentry. The atmosphere predictably incinerated the ship’s engines and exposed steel skin.

If this afterthought feels like a conspiracy theory, remember the time when Musk made a change to starship after Tim Dodd (earlyastronaut) asked him a question on the same? Or the time when someone on Twitter asked Musk why they didn’t start two raptor engines and then shutoff the underperforming one during Starship’s flip maneuver at landing? They do this now. Afterthoughts are evidently not a rare thing at SpaceX.

And finally flight 4. I’m not taking any credit away from them. They seem to have just made it till the landing. Superheavy worked all the way for the first time. But my concern is about the place where the starship’s fin burned through - exactly at the hinge. I would have expected them to focus more on that region as a weak spot and to have given it a better thermal protection. That would be the last region I would expect a burn through. Instead, it would have been in some spot where they didn’t expect any problem and missed something very subtle.

All these give me the impression that they are trying things and seeing what sticks. That’s not how traditional rocketry works. There’s a s**t load of analyses, simulations and small scale tests that precede the production stage. The result is that when such rockets fail, they fail in a spectacularly complex, unpredictable and mind bending sequence of cascading failures (unlike what I see on starship). They also tend to succeed with minimum test flights and work reliably over decades. Apollo is a great example. The first test flight achieved everything that starship achieved in 4 flights.

The only other industry that I’ve seen behaving like this is the IT industry. “Deploy whatever you have and we’ll debug in production”. Coincidentally, Tesla does the same with their cars - the only car company to do so. So perhaps there is a Musk factor here. The SpaceX engineering team is incredibly talented, skilled and capable. The only reason I can think of for them to behave like this is an enormous pressure on them to deliver results at high rates. That’s the only reason I can think of for them to proceed without satisfying themselves.

Now you may want to argue that SpaceX’s approach is better than what everyone else does. After all, they make bigger things, faster. They advanced the industry like no one else did or could. Perhaps you’re right. Only time will tell, since this approach is so novel that we haven’t had enough opportunity to assess the results. But my instincts worry me about one thing - technical debt. Mechanical engineering is not like software engineering. In software, a problem once solved is gone forever. In mechanical engineering, any problem you work around is a disaster waiting to happen in the future. The right approach here is to design things properly and meticulously so that the final product has minimum work around. I fear that the software style of design is leaving unknown flaws or technical debts that may compound together into a cascading failure on some flight in the future.

To those who are planning to reply:

What I wrote is not a criticism of the SpaceX employees or Starship programme. I’m genuinely fascinated by the interplay of technical design, development styles, management styles and human factors. I’m extremely curious about how the situation is evolving.

Please don’t attack me or question my abilities if you feel that I’m being unfair about this (that’s definitely not my intention). I may be just a kid in this arena, but it’s never wrong to ask, is it? I’m extremely interested in hearing your insightful opinion on this topic, based on your experience or your logic. If you think I’m wrong on any of these, please share your perspective and reasons here - I (and possibly others) may learn something new.

I’m sure that this won’t stop the idiots who argue in support of using Discord as a support channel for FOSS projects.

Let me know if and when this makes insulin cheap enough to afford. If we’re going to continue making big companies richer at the expense of sick people, we might as well not gloat about these achievements.

And if you’re going to talk about the dependence of price on demand and supply, you’re still not getting it. These companies are masters at creating artificial scarcity by several means including patents and price gouging cartels.

Imagine the scenario where you have to bribe (disguised as a subscription) each megatech company to respect your privacy. How many times and how much will you be willing to pay for something that should be your fundamental right?

Given Meta’s history, no one should misinterpret their intentions. They should be outright banned for these egregious transgressions.

My point is that it is not correct that only gmail works.

Proton Mail and Fastmail are two such examples

I had Proton and Fastmail in mind when I wrote that reply - along with some others like Migadu. But the point remains - you need to be a large scale mail provider like any of them to bargain your way into deliverability. Self hosting is completely left out - a far cry from the actual federated design of emails.

It is a lot of work and fairly costly to get correctly setup and a real pain to maintain.

That is not necessarily true these days. There are a lot of turnkey bundles like mailinabox, mailu and mailcow that are easy to deploy, maintain and update. There are even projects that aim to combine all the necessary services into a single binary server - like maddy and stalwart. They provide everything necessary for running a mail server and ensuring deliverability. But they still don’t get delivered on those large service providers.

There is very little reason to do it too

Paid services that don’t sell you out is much better than big free ones that squeeze you for data. But I don’t consider having our data on someone else’s server as ideal. The ideal thing is that every home should have a cheap server with net/web applications that can be deployed with the ease of installing a desktop application. But we are going the wrong way. And those big monolithic abusive service providers are the biggest hurdle to achieving that.

‘It works for me’ isn’t an answer and that doesn’t make it overblown. This is a well known problem for years.

I applaud your intention. But there really is such a thing as peer pressure. The reason why only Gmail/GSuite and Hotmail/365mail is left is because these two ensured that mail wasn’t deliverable from independent email servers. And the fact that a vast majority was on them helped them convert a federated messaging medium into more-or-less centralized service. Even today, there are a lot of people around who knows the harm in using Chrome, but then goes ahead and say - “I would have switched to Firefox, if only some-random-useless-website worked on it”. It’s always possible for people to harass the company/institution to support Firefox. But they would rather make up excuses to stay on Chrome than do something about it. The same happens to every Google service as well - especially Gmail.

Sure. The Japanese seem to have a lot of issues with nozzles.

If what you say is true, this is the second time a Japanese mission has been salvaged after a failure of rocket nozzle. The Kaguya mission to Venus failed to enter Venusian orbit after the main engine nozzle exploded. They managed to enter the orbit during their next pass using the smaller attitude thrusters.

deleted by creator

Like somebody else here said, it isn’t because these magics aren’t viable. It’s just more profitable to fleece sick people with treatments than it is to cure them.

Surprisingly, my guess was C too. And for some reason (which I’m trying to recollect), my mental model was almost exactly the same as in the parallel paths experiment - I was expecting the currents to ‘slosh around’. So it wasn’t a big surprise when the initial probe showed multiple steps after switching.

I guess there are multiple ideas at play here. The first is that it’s not very accurate to model current as the flow of electrons. Current is more like a wave in a sea of electrons. Meaning that though current moves at speeds approaching the speed of light, the electrons themselves are much slower. This is what makes the water channel model in the video so appropriate.

The second is about characteristic impedance. It isn’t just another impedance. It’s a point property of the transmission line. It’s also fundamentally related to wave propagation. It comes up anywhere wave propagation is involved - for example mediums that carry sound waves have acoustic characteristic impedance. An intuitive explanation of the concept…may take a few pages. I just don’t have the energy to do it now. I will just say that it’s a really really important property. It’s well worth it to go back to it, study and iron out all the misconceptions, even if it takes you hours to do it.

A lot of stupid techno wannabes will think that this is cool and ruin it for everyone else. We need that laughing man tech from Ghost in a Shell.

Looks like they did a skip at 4:00 with a bit of aerobraking before the actual reentry. That would make sense, since the velocity is incredible (11 kmps), coming straight from the moon. Velocities are lesser in case or reentry from LEO. Two-phase entry might reduce the load on the reentry shield and its occupants.

22:58 - attitude control kick in again while under shuts. Why ??

Just speculating. There isn’t much of an attitude change except in the initial few seconds, considering the amount of firing it does. It’s probably intended to consume the propellants and render the capsule safe. If that’s indeed what happened, they were likely firing opposing thrusters (in combo or alternatively), cancelling it out. There is such an operation in space tech - called passivation. It’s the space equivalent of a fuel dump.

PS: If anybody is wondering - this isn’t a pure ballistic reentry. Modern spacecrafts, even the capsule shaped ones, can change their trajectory by rolling on their axis of symmetry (the axis that passes through the center of the heat shield and top of the spacecraft). This may sound counterintuitive, but it has been used so many times already - like in case of the Mars rovers. They offset the C.G of the spacecraft a little bit away from the axis and uses it asymmetry to achieve guidance.

It’s true that most people won’t know or care. But the only ones who can make a difference are the ones who understand the situation. We shouldn’t assume that nobody will listen to us. If we tell a hundred people, perhaps 5 will listen - and even that’s a pessimistic assumption. Even such small changes add up in the long run. The last nail on the coffin of our freedom will be the silence of those with the wisdom to recognize its erosion.

Google has also started delaying the approval of revisions of privacy-related addons. This is an all out war against user privacy. Everyone, please stop using or promoting this Trojan malware called chrome or anything based on it.

I’m not a doctor, but a technologist. A new diagnostics tool? Good! But I can guarantee that good doctors won’t ditch the stethoscope for anything else. No amount of automation can replace the reassurance that your own senses give you.

This is perhaps more true in the case of mechanical engineering. The touch and feel of the machines can be unfamiliar and changing. And the diagnostics tools are more available (because invasive probing is more acceptable). Despite this, I have noticed that myself and others depend on sight, sounds, heat and sometimes even smell to keep a tab on their health. I don’t think any professional would consider completely cutting off their own senses.

These companies dig up everything from a prospective employee’s past, starting from childhood. Things that you said come back to bite you, even if you got wiser and changed your stance. But companies get to pull shit like this without consequences.

There should be a public blacklist database with every company and their dirty infractions like these. That way, at least very competent and desirable candidates can avoid them and look for better jobs.

I don’t have anything to hide

Great! Then I guess they don’t mind giving you their bank password, credit card pin, details of all the medicines they take, information from the work they do, their detailed weekly activity schedule, their browser history, their investment portfolio and assets, etc, etc… I’m salivating at the thought of the hundreds of different ways in which I can make money with all that info!

Email hosting is hard for two reasons. The first is that there are too many parts to configure - MTA, MDA, DKIM, RDNS, spam filter, webmail, etc. The viable solution is to use a turnkey solution like mailinabox, mailcow or mailu.

The second problem is deliverability. At the minimum, you will have to ‘warm up’ the server. You will have to send a few dozen mails to others and ask them to mark as not-spam. Even then, a lot of other factors come into play - like the IP address block (for example, mails from AWS always gets blocked), domain name and even the top-level domain - they all influence the spam filter score.

Meanwhile, deliverability with Google and Microsoft (incl google workspace and ms 365) are lost causes. Google sends your mail to the spam folder irrespective of your spamassasin score. They provide no viable solution to this. MS on the other hand just drops mail silently. This isn’t a bug. Both of them are trying to destroy the federated nature of email and consolidate all email business to themselves.

Meanwhile, the big players like fastmail and migadu get better treatment. Especially, migadu is a good choice if you want unlimited aliases.

Finally, talking about aliases. Most services (except migadu) offer only a few aliases. That limitation is not there for selfhosted email. An alternative to aliases is to use + addresses (eg: mybox+bank@mydomain.com). The advantage of this method is that you can make up multiple addresses on the fly (without registering) using a single alias/address. You can use this in combination with a filter like sieve (server-side) or notmuch (client-side) to sort and filter incoming mail.