Hacker News Reader

Comment by npalli | Hacker News Reader

No, his attitude is entirely appropriate. Learn your basics before you try to reinvent stuff. As Alan Kay keeps reminding us, the programming community doesn't really learn anything that is done in the past and is doomed to keep reinventing stuff or working on things that have been shown not to work. For example, if you study physics you are not going to go far if you show up to class and start dropping apples and say this shows some magic force is operating. Learn the basics of engineering first then try to change things. Otherwise you will just waste everybodys time.

As for your programming example, the more appropriate case would be if someone showed up with a revolutionary proposal to allocate memory for a GC, you see the source code and there is a big array structure defined locally "char memory[10000000]". There is far more things wrong here than trying to figure out ways to scale this in production.

View on HN

"Learn the basics of engineering first then try to change things. Otherwise you will just waste everybodys time."

So what about SpaceX and Tesla? To me those indicate that Musk must be a pretty decent engineer, and perhaps more importantly, must be excellent at leading real hardcore engineers in complex domains.

Tesla and SpaceX are great but I'm not aware of fundamental engineering principles they have overthrown. This is a strange religious argument you are making here. The logical conclusion to your attitude is that there is no engineering challenge that Musk and his compadres can't solve. How odd. People have been building bridges for millenia now, there is a rich history of engineering and science here. You can't just 'think out of the box' your way through some fundamental constraints.

If you feel the original article is wrong then it is worth discussing. If Musk and friends are modern day messiahs what's there to discuss?

>Tesla and SpaceX are great but I'm not aware of fundamental engineering principles they have overthrown.

No, but it is a pretty good claim to show an understanding of the basics that they supposedly haven't learned.

How so? The point being made in the blog post is that Musk has a severe misunderstanding of the fundamentals of civil engineering.

That Musk runs a company that has successfully engineered cars and spacecraft in no way makes him qualified to construct the Hyperloop.

Civil engineering isn't mechanical engineering isn't chemical engineering isn't aerospace engineering. These are different disciplines. Being a talented industrialist in one (or more) of these do not imply capability in any of the others.

> Civil engineering isn't mechanical engineering isn't chemical engineering isn't aerospace engineering.

That's true, but all of them derive from the same set of first principles, those defined by physics.

> Being a talented industrialist in one (or more) of these do not imply capability in any of the others.

Not true -- expertise is not implied, but capability certainly is. In this specific case, to predict the behavior of an evacuated transport tube, and given that the project doesn't resemble any existing projects, one could hardly do better than have a background in spacecraft booster design.

I can't disagree more. I'm a mechanical engineering by schooling (what am I doing writing code?!) and it still routinely shocks me the sort of intriguing misconceptions I hear coming out of the mouths of very smart, very technical people.

The fact that mechanical engineering and civil engineering and aerospace engineering all derive from basic physical laws is rather meaningless - engineering isn't about first principles, it's about the extremely complex task of putting these to use.

To make an analogy to software, that's like saying "writing code for a Martian probe isn't very different from writing code for a website, it's all the same first principles".

In this case, the author's thesis is that Musk is applying analysis methods that are wholly inapplicable to civil engineering, and fails to account to the scale of the structure, and makes faulty assumptions about the validity of simulating the structure in small segments. This is the sort of thing you see when you cross someone who is an expert in one field into another where they have no expertise - but they know just enough to get started.

I write mobile apps for a living. I can probably hold a decent conversation with a kernel hacker, but I am in no way qualified to design the next uber-kernel of the future. Traditional engineering is no different.

> I'm a mechanical engineering by schooling (what am I doing writing code?!) and it still routinely shocks me the sort of intriguing misconceptions I hear coming out of the mouths of very smart, very technical people.

That's true, and I've had similar experiences, but it doesn't argue against my saying that these engineering disciplines have many basics in common.

My only reason for posting was to argue against the idea that different engineering fields have no common base. They either do, or they're not based in science, which is the common base for all engineering fields.

As to a new project that has little in common with existing projects or established engineering fields, the obvious solution is to depend on basic science rather than nonexistent engineering projects, which would be the best approach to a project like this one.

All new engineering projects teach us one thing -- no matter how many advance expectations and predictions, by the time the project is operational, all of us will be educated in new principles that no one anticipated.

For me the various engineering streams are just as different from each other as from every other non-engineering discipline. There is no way a mechanical engineer could work in the electronics, IT or biochemical engineering fields.

If you are going to label the common base as "science" then you could equally say engineering is common with psychology since they both involve "words". That makes equally as much sense.

> For me the various engineering streams are just as different from each other as from every other non-engineering discipline.

False, and that cannot be used to argue that they're not all directly derived from physics. Look at cosmology and particle physics. It's hard to imagine two fields that are less alike -- one studies nature at the largest scales, the other at the smallest. But not only are they both physical fields and share the same theories, but they attend each other's conferences and work together on common issues.

Remember that engineering is applied science. This is true for all engineering fields in which people place their trust.

> There is no way a mechanical engineer could work in the electronics, IT or biochemical engineering fields.

Same reply -- it's not relevant to the issue of their scientific underpinnings.

> If you are going to label the common base as "science" then you could equally say engineering is common with psychology since they both involve "words".

When you think of a coherent argument, post again. Psychology isn't scientific, engineering is, and the topic is science. When a psychologist uses the word "gravity", he means seriousness. When an engineer uses the word, he means spacetime curvature.

> That's true, but all of them derive from the same set of first principles, those defined by physics.

More is different (PDF link: http://robotics.cs.tamu.edu/dshell/cs689/papers/anderson72mo... ).

Much like a knowledge of physics is insufficient to draw conclusions about biochemistry, knowledge of basic physical laws tells you very little about the fundamentals of a practical engineering field. The theory and laws of e.g. civil engineering are typically traceable back to abstract physical principles, but are not strictly derivable from them.

> More is different (PDF link: http://robotics.cs.tamu.edu/dshell/cs689/papers/anderson72mo.... ).

What? A link to a paywalled article with no abstract? That was constructive. Here's a real link:

https://www.tkm.kit.edu/downloads/TKM1_2011_more_is_differen...

As to the article's content, it's philosophy without supporting evidence (meaning it's philosophy), and its conclusions are regularly refuted by the many examples that connect physical fundamentals with everyday reality. Example? Quantum theory and computers -- as separate as the two may seem, we can't have the second without a detailed knowledge of the first.

> Much like a knowledge of physics is insufficient to draw conclusions about biochemistry ...

In the final analysis, all of biochemistry derives from biology and chemistry, each of which derives from physics. Relevant in that regard is the fact that chemistry turns out to be applied quantum physics, to a shocking number of decimal places.

> The theory and laws of e.g. civil engineering are typically traceable back to abstract physical principles, but are not strictly derivable from them.

Not true. Either civil engineering is "strictly derivable" from physics, or it's not part of reality. The behavior of concrete over time, as just one example -- we know what will become of it in the long term, and modern quantum theory tells us why.

Philosophical articles like "More Is Different" represent nothing more or less than physics envy. They're interesting and deserve to be heard, but in the reality footrace, events annoy the authors by reliably overtaking their philosophical positions.

There is a difference between reality and academic physics. It is trivial to say engineering knowledge is derived from reality, and it's wrong to say it's derived/derivable from purely theoretical/academic knowledge of physics -- especially because good engineering requires knowledge of man-made safety standards and such.

Engineering is a corpus of knowledge that has been slowly and with great difficulty hewed from the raw rock of real experience over hundreds of years. It is not just a branch of math derivable from axioms. Show some respect.

> There is a difference between reality and academic physics.

"Academic physics" describes reality. It is the closest thing to an accurate description of reality that we have.

http://en.wikipedia.org/wiki/Theoretical_physics

Quote: "A physical theory is a model of physical events. It is judged by the extent to which its predictions agree with empirical observations."

Read the above carefully. What it says in essence is that a physical theory is judged by the degree to which it agrees with reality. Our best physical theories agree with reality to ten decimal places.

Physical theory is a statement about reality. If reality disagrees, the theory is thrown out and replaced by a better one.

> It is trivial to say engineering knowledge is derived from reality, and it's wrong to say it's derived/derivable from purely theoretical/academic knowledge of physics ...

Engineering is applied science. Any engineering field for which this is not true has no right to exist, and doesn't deserve the public's trust.

http://en.wikipedia.org/wiki/Applied_science

Quote: "Applied science is typically engineering, which develops technology, although there might be feedback between basic science and applied science: research and development (R&D)."

> Engineering is a corpus of knowledge that has been slowly and with great difficulty hewed from the raw rock of real experience over hundreds of years. It is not just a branch of math derivable from axioms.

False, and you need to learn this topic. I hope you're not an engineer.

> Show some respect.

By correctly identifying engineering as applied science, I did. There is no higher praise than to say -- correctly -- that engineering is derived from scientific principles, from physics.

No offense, but I think you are seriously underestimating several domains of knowledge.

> No offense, but I think you are seriously underestimating several domains of knowledge.

How so? By insisting on their connection with established scientific principles? Astrologers would love for me to do that, as would psychologists.

How, then, do you account for the very basic and substantial errors described in the article?

> How, then, do you account for the very basic and substantial errors described in the article?

Surely you don't think a single counterexample -- or a dozen counterexamples -- refute well-established principles? The argument was that various engineering disciples are independent of each other and of physics. That's very clearly false and smacks of post-modernism (i.e. that everything is a matter of opinion, not evidence), and it's hardly supported by an example -- or a dozen examples -- where someone gets the engineering wrong.

> The argument was that various engineering disciples are independent of each other and of physics.

No, actually, they were argued to be different than each other, but no one in this thread argued that they were independent of each other or that any of them was independent of physics.

Inasmuch as that's the position you are arguing against, it is a strawman.

>> The argument was that various engineering disciples are independent of each other and of physics.

> No, actually, they were argued to be different than each other, but no one in this thread argued that they were independent of each other or that any of them was independent of physics.

False. Read this quote: "The fact that mechanical engineering and civil engineering and aerospace engineering all derive from basic physical laws is rather meaningless - engineering isn't about first principles ..."

To say, as this correspondent did, that "engineering isn't about first principles", is to argue that it has nothing to do with science, because science establishes first principles.

> Inasmuch as that's the position you are arguing against, it is a strawman.

Well, since you can't be bothered to read the thread and find out what positions people have taken and that I'm arguing against, there's no point to this.

> To say, as this correspondent did, that "engineering isn't about first principles", is to argue that it has nothing to do with science

It is not arguing, as you claimed, that the disciplines or "independent of each other and of physics", especially not when, as that correspondent did, an acknowledgement is included in the same sentence which expressly acknowledges the fact (while dismissing its relevance in context) that they all derive from the same physical laws.

No one is arguing that different engineering disciplines aren't related to each other and to physics; the debate is about the extent to which how those relations (as well as Musk's connection to certain engineering disciplines through SpaceX and Tesla) are relevant in the context of the Hyperloop proposal, the criticism of it, and the treatment of Musk as an authority deserving special deference in regard to that proposal.

I was addressing your claim about this specific case:

"In this specific case, to predict the behavior of an evacuated transport tube, and given that the project doesn't resemble any existing projects, one could hardly do better than have a background in spacecraft booster design."

Seems to me that one could do a lot better than having a background in spacecraft booster design, as the fairly trivial errors pointed out in this article illustrate.

His primary argument is about thermal expansion. Are you claiming aerospace and automotive engineers make no study of thermal expansion? I could see your point if the proposal was about a dirt retaining wall or aqueduct where the other two disciplines wouldn't be well versed in the material's properties, but I am not seeing something CE specific in the thermal expansion of steel.

Yes, “stringer” is also a common word in structural engineering, especially in bridge design, but something that reinforces a tube wall is a stiffener, not a stringer. Update: I am reminded by Matthew Hunt on Twitter that “stringer” is commonly used in aircraft design. I should have remembered that.

Given this quote how do we know they aren't in violent agreement but talking past each other because they are using terms from different disciplines?

At least SpaceX actually made some quite fundamental changes in their approach to many things, since their original ideas did not work. They for example talked a lot about recovering their rockets, and none of that has worked at all. They changed from ablative to regenerative cooling. They experienced many failures (I think it is actually better to launch more often and experience some failures than launch years apart, because then you can't afford to fail at any cost and your schedule and cost will explode).

So in a sense he has this "new kid on the block" serial enterpreneur attitude where he bursts into a scene, stumbles a lot but finally seems to get something done quite right. Maybe it's him learning the tricks of the trade - or that just finding the right people takes a lot of time, or something else entirely. I at least don't know that closely what happens in these organizations.

It might be cool to read a book about something like that. I don't have high hopes of anyone getting the stories right though.

What's your point? Because you're good at something doesn't mean you're good at everything.

That being said, I agree that the tone of the article is a bit too harsh for my taste, but then again Linux Torvalds is revered here for this kind of tone. It's always good to think outside of the box and bring new ideas to the table. Even if they're dead ends, they keep the brains working, if only to form counter proposals.

N.B.: american friends, pretty please, consider including metric values alongside imperial, I have no idea what 1030feet means. Fortunately we have google for unit conversion...

A foot is about a third of meter, and a mile is about a kilometer and a half. That's all you really need to remember. Heck, in a discussion like this you could pretend a mile equals a kilometer and be fine.

Oh, and a pound is about half a kilogram. I won't expect you to remember all the stupid volume units, but those three tiny conversions are worth knowing.

And a liter is about 1/4 of a gallon (useful when calculating gasoline prices.)

Or about 2 pints, useful for beer.

> Because you're good at something doesn't mean you're good at everything.

How is that statement relevant here? The criticism being responded to was that he doesn't know the basics of engineering. The response was there's significant evidence he knows a lot about engineering (and his employees who looked at this proposal probably know more). It doesn't matter whether he's good at everything.

edit: fixed typo and added emphasis

> How is that statement relevant here?

Because in real life (as opposed to, say, Star Trek), aerospace engineering and civil engineering ain't the same ballpark, ain't the same league, and ain't even the same flipping sport. [1]

[1] with apologies to Pulp Fiction.

Neither are online payments, electric cars and aerospace engineering similar sports. His core competence seems to be learning new sports very quickly.

He may strike out, he may not. If he fails it's not going to be because his first paper over-reached or because he's unable to gather smart-enough experts with better civeng software.

> If he fails

So far, its not clear that he is attempting anything with Hyperloop other than what he is demonstrated to be very good at: selling ideas.

Its particularly clear, though, that the one thing he isn't trying to sell Hyperloop as yet (and may not be at all) is an actual project/business, as opposed to just a reason to not support CA HSR.

Why the CEO of a company selling currently selling high-end electric cars and working to expand the reach of electric cars in the market might want to sell an alternative to a project designed primarily to reduce automobile road trips (which incidentally reduces the need for automobiles) in California by improving mass transit on both the major intrastate long-distance corridor and the connecting regional/commuter mass transit systems -- either as a real system that doesn't do as much to reduce automobile demand (and even relies on transporting autos to provide door-to-door connectivity) or as a phantom system to derail the existing proposal -- is pretty obvious.

Hyperloop doesn't need to be successful as mass transit -- or even ever have $1 spent toward building it -- to succeed for Musk.

> its not clear that he is attempting anything with Hyperloop other than what he is demonstrated to be very good at: selling ideas.

While selling ideas isn't a bad skill I think you're underselling his ability to execute, somewhat.

Personally, I feel that Musk's "list of biggest risks to Tesla" has many items before "CA HSR". By the time it's built Tesla will either be (at least) a countrywide success, or not.

> While selling ideas isn't a bad skill I think you're underselling his ability to execute, somewhat.

Not my intent; I think SpaceX and Tesla both show Musk can execute in engineering quite well when his energy is behind it (or, equivalently, that he is good at putting his energy in places where he can execute.)

Part of that is the ability to sell ideas to (among others) investors, without which he wouldn't have the ability to execute.

With regard to hyperloop, Musk has been very visibly not putting the kind of intent-to-execute behind it that he has with other endeavors, making his selling ideas skill the most relevant part of his ability to execute, but I certainly did not mean to fail to recognize his ability to execute.

> Personally, I feel that Musk's "list of biggest risks to Tesla" has many items before "CA HSR".

Probably. But tossing out a "I don't plan to build this, but it would be better" diversion is a pretty low-cost way of derailing the threat. It doesn't have to be the biggest or most immediate threat to be worthy of that, from a rational viewpoint.

> By the time it's built Tesla will either be (at least) a countrywide success, or not.

This presumes HSR (including the regional/commuter transit improvements that go with that project, and are some of the earlier investments) has no effect on the market for electric cars until its fully built out. I'd question that assumption.

No, but a lot of the same types of problems surface even in all these different fields, engineering does have some rather uniquely diversified fields, but at their core they still follow ...a lot of... the same laws of physics.

arethuza ninja-edited his comment after my reply, his initial reply was simply "So what about SpaceX and Tesla?", which didn't make a lot of sense to me.

But my point still holds, because you're good at making rockets and electric carcs doesn't automatically make you an expert in high speed train infrastructure, so I don't really think that's a valid argument.

It means you know a whole lot about thermal expansion.

It might mean you should know that, but to the extent it does, that makes it more, not less, notable that your work fails to demonstrate that.

Okay, but who is Dr. Drang? As nearly as I can tell from a web search, he is someone puts some code on Github and writes a blog. Has he sent rockets into space? Why does he feel like he gets to talk down to someone who has sent rockets into space, on the subject of engineering?

Now that's an appeal to authority. Regardless of whether this person is a doctor or a billionaire or a physicist, the key points (like the issues with thermal expansion) are legitimate criticisms. "Dr Drang" presented arguments that are within the reach of high school physics.

Now, Musk may have an answer to the issues, but he certainly didn't present them. The fact that there is no obvious resolution suggests that either Musk didn't think of them or he intentionally omitted the answer (both are damning)

> key points (like the issues with thermal expansion) are legitimate criticisms

No, thermal expansion criticism is not careful enough. See my comment:

https://news.ycombinator.com/item?id=6238022

Yeah, this was my initial confusion with the article and many of the comments. When I read the doc by Musk, I assumed, that the tube 'floated' over the pylons. The author seems to have assumed the opposite. The problem is that they are both assumptions.

When making an argument, sometimes it is good to make an assumption and then prove the assumption is false... sometimes it is just a straw man.

The tube does float (longitudinally) over the pylons, in the proposal.

The tube will be supported by pillars which constrain the tube in the vertical direction but allow longitudinal slip for thermal expansion as well as dampened lateral slip to reduce the risk posed by earthquakes. In addition, the pillar to tube connection nominal position will be adjustable vertically and laterally to ensure proper alignment despite possible ground settling. These minimally constrained pillars to tube joints will also allow a smoother ride.

> As nearly as I can tell from a web search, he is someone puts some code on Github and writes a blog.

A civil and mechanical consulting engineer [1] who also has put some code on Github and writes a blog.

> Why does he feel like he gets to talk down to someone who has sent rockets into space, on the subject of engineering?

Probably because, unlike Musk, he actually has professional qualifications in the kind of engineering at issue on the points on which he is criticizing Musk's proposal.

[1] See, e.g., http://www.codeproject.com/Articles/582981/A-Coder-Interview...

"Dr. Drang, a consulting engineer well known amongst nerds on the Internet" (from http://5by5.tv/systematic/57)

Seems super legit

1 foot ~= 1 nano-light-second

Tesla and SpaceX are both iterative technology. They both use existing technology and iterate improvements on it or apply it to new fields. The electric car was already proven feasible by GM and other companies or products like golf carts. It was more of a problem of the existing players being forced to build an electric car than keep doing what they've been doing to generate profits.

SpaceX still uses rockets. The new tech in SpaceX is changing the rocket design between engines so it is manufacturable in larger quantities and utilizing faster computers to create reusable rocket stages that can land with their own power back on Earth. You could say that SpaceX is leveraging computer science artificial intelligence research, the general improvement in computing (cost and speed), and sensors. For example there are many videos of software controlled quadcopters performing various aerobatic stunts by analyzing and adjusting thrust in realtime based on sensor information.

One could say there is nothing new in feedback control for vertical flight. It has even been coupled with GPS for precision motion and landing in rockets already many years ago by Armadillo Aerospace and others.

All this could have been done in the seventies, bar the GPS part, which would have required some more costly alternatives.

But even the V-2 had gyros on a free rotating platform that was connected to the thrust vectoring vanes that kept the rocket pointed to where the flight program required. The sensors weren't so good and you probably could not have throttled the V-2 to hover accurately, but maybe with radio guidance (that they actually used with some of the flights) it could have worked. Mechanical PID controllers date to late 1800s already (warship steering) so it was a totally straightforward thing to do, and was done by Goddard already. Nobody cared but the Germans...

There is very little that is novel about SpaceX / Tesla. Both were excellent executions of visions that many people had. We've had pure-electric vehicles for more than a century (For example, Detroit Electric built an electric vehicle in 1914 -- https://upload.wikimedia.org/wikipedia/commons/8/8a/EdisonEl... is a photo with Thomas Edison).

Hyperloop is different. Others have talked about it, but this is the first time he is truly laying out an idea, so I wouldn't necessarily assume that he will be successful

> So what about SpaceX and Tesla? To me those indicate that Musk must be a pretty decent engineer, and perhaps more importantly, must be excellent at leading real hardcore engineers in complex domains.

Well, then, how do you explain the following?

  1- Thermal expansion and contraction issues diametrically 
     opposed to what happens in the real world
  
  2- Proposing that an entire 300 or 400 mile tube be a single 
     welded solid tube with no turns or incremental expansion
     gaps.  Really?  In the real world?  Along the 5 freeway?

  3- Grotesquely underestimating seismic effects
  
  4- Estimating a cost at least one order of magnitude 
     lower than what it will actually be in the real world
  
  5- Not taking a drive down the Golden State highway before proposing
     that columns be installed every 100 feet to support the tubes.
     No, really, go on Google Maps and take the drive yourself.

Start here: http://goo.gl/maps/zPPKa

I think these guys thought it would be a cool and possibly marketing-worthy thing to put out something like this. They put a little time into it here and there and, per the paper, a short amount of full time effort. They were probably so focused on the tube transport physics that they neglected to connect to the real world.

At some level I don't fault them for putting this out. They are obviously far more focused on building cars and rockets and that's exactly how it should be. If what came out was a fully engineered project it would have meant that they took valuable time away from Tesla and SpaceX to apply on this thing. That could border on being a breach of fiduciary duty.

At this point I am going to mark this off as science fiction simply because I am now convinced it cannot be built outside of a super-expensive short range test installation.

Why would you assume that the founder of SpaceX and Tesla "must be a pretty decent engineer"? I'm not saying Musk isn't, because I don't know much about him, but I certainly wouldn't assume that the founder of engineering-based companies must be a decent engineer. As for leading engineers in complex domains, that's more plausible, although the founder of a company doesn't need to be the person directly managing engineering teams.

Exactly. Give me enough money and I could easily build a SpaceX and Tesla.

How ? By hiring people who know how to easily build a SpaceX and Tesla.

If you're wondering where the conversation got derailed it was here:

> Learn the basics of engineering

Every single comment resulting from this comment has been partisan:

Side 1: Elon Musk knows the basics of engineering as evidenced by his numerous successful projects.

Side 2: Strawman: although the original comment was about _basics of engineering_ turn it into an argument about the differences between engineering disciplines.

Seriously folks, if you're going to argue about semantics, skip the strawman, he ruins the party.

> So what about SpaceX and Tesla?

Neither of those depend on the kind of engineering necessary for a large terrestrial infrastructure project (and the required engineering is more similar between SpaceX and Tesla than between either and a large civil engineering project.)

If I recall correctly, Tesla had difficulty designing a transmission at first go and they are in the car business.

So I guess the point is, apparently persistence counts for something.

A static allocation works perfectly well. If we want a hyperloop analogy, maybe they specifically pointed out that the runtime is 3x faster if you don't have to allocate during it, and that's what they're optimizing for. Nobody cares if it's revolutionary.

On the other hand, if you're going to mock a proposal, it makes good sense to actually read it thoroughly. His criticisms on the thermal expansion issue are founded on the mistaken assumption that the tube is fixed to the pylons; this is contradicted by the first paragraph in the pylon section, which indicates that the pipe is not fixed longitudinally to the pylons. It slides.

Reply on news.ycombinator.com