Against the Hydraulic Analogy | Hacker News Reader

Against the Hydraulic Analogy

lcamtuf.substack.com

47 pointssurprisetalk33 comments

This seems unnecessarily negative. Teaching analogies don't need to perfectly resemble the system being taught. From a pedagogical perspective, the aim is typically to relate something unfamiliar to something familiar, like water flowing in a pipe, in order to help someone gain footing in a topic.

Further connections can be developed mathematically; electric force propagation does obey wave equations, after all. But by then the analogy has served its purpose and should be droppped in favor of more rigorous knowledge development.

When I was in EE at Purdue, most of my profs also railed against the hydraulic model. It really messes you up when it comes time to get into higher frequency digital circuits (CPU designs were just starting to brush up against quantum problems at the time).

Why would anyone be attempting to use this analogy to learn about intermediate to advanced topics like that? That's not what learning aid analogies are for... They are for beginners to gain some initial intuition in order to get over the immediate hump of unfamiliarity with a new subject.

They wouldn't, but people would create these increasingly tortured analogies out of the same motivations that other creative people would create, say, a raytracer in Brainfuck or a pregnancy test that can run DOOM.

By the time you get into higher frequency circuits I would think you would already understand exactly where the hydraulic analogy breaks down. Most grade school students aren’t doing any kind of extremely high frequency work, let alone even basic 60hz work.

When I was taking undergrad EE classes, I don’t recall it ever coming up. I think people’s comfort with and intuition for electricity are at least as good as for water, which actually behaves in deeply unintuitive ways. We’ve been living with switches and toaster ovens for generations now, people understand resistance, voltage, and current just fine on their own terms. Or at least, water analogies don’t help because people don’t understand water any better.

Once you start dealing with EM fields and even fairly common filter designs in the AC domain, the hydraulic analogy starts breaking down and becomes an impedement to learning... so my prof would say.

I was never taught the water analogy but I saw quite a few students struggle trying to wrap their heads around a physics (EM, RF, potentials, etc) based circuit design and theory.

As a fellow EE, I was amused that the plumbing model never came up in my engineering classes, but was introduced (and re-introduced) at several levels in my physics classes.

Agree. The hydraulic analogy is a good introduction to the duality that is current and voltage — as well as, to varying degree, capacitance and resistance. But then let's just stop there.

No problem.

I completely agree. Every model is "wrong" to some degree, it's just a matter of applying the "least wrong" one in the proper context to yield the most practicable solution.

Even basic circuit theory (Kirchhoff's laws) falls apart when the preponderance of propagating wavelengths become close to the characteristic length of the propagating medium.

Then that "wrong" simplifying model must transition to the much more complex transmission line theory.

By the same analogy, one doesn't use basic Bernoulli pressure/velocity/flow equations for plumbing when dealing with chaordic or turbulent flows outside of very limited Reynolds number regimes--one has to do numeric simulations of the more general Naiver-Stokes vector differential equations.

>Wikipedia doesn’t stop there; here’s their take on the metal-oxide-semiconductor field-effect transistor (MOSFET):

This stood out to me because I have contributed quite a bit to the Wikipedia entry for MOSFET.

This illustration does not appear on the Wikipedia entry for a MOSFET: https://en.wikipedia.org/wiki/MOSFET

It only appears on the Wikipedia entry for "hydraulic analogy". People do not learn electronics from the Wikipedia entry for "hydraulic analogy".

I assert that it is highly likely that not a single human being on earth has ever used the Wikipedia entry for "hydraulic analogy" to learn electronics. It is likely that several people have used that entry to learn about the hydraulic analogy.

Then again, I learned electronics from 1988's The Way Things Work, which uses Wooly Mammoths, large stone edifices, and titanically large humans so maybe I was trained on analogy from an early age and learned to not become too attached to them when they start to break down (like how Wooly Mammoths don't exist anymore).

edit: oh goodness I just looked it up and the hydraulic analogy for a circuit in The Way Things Work is utterly magnificent, fantabulous, funny, and educational. No mammoths, unfortunately.

> 1988's The Way Things Work

I'm so glad to see this book mentioned on here! It was my favorite as a kid and it still holds up today IMO. I have one of the newer editions at home and this comment reminded me to go page through it a bit. Still brings a smile to my face.

https://en.wikipedia.org/wiki/The_Way_Things_Work

Exactly! It's a wonderful Lie to Children.

https://en.wikipedia.org/wiki/Lie-to-children

Just because the way you understand a profound truth is not the same way it was first introduced to you does not mean that the first introduction was flawed or wrong. It, more likely than not, means you're less of a child now than you were then.

I found a copy of that book in excellent condition at a thrift store a year or so ago. Paid $1. One of the best bargains I've scored in a long time.

My problem with this "hydraulic" analogy (first time I've seen it called that) is that I feel most people don't have that good intuition on hydraulics either, so you are just trading one poorly understood subject with another almost as poorly understood one.

Maybe 50 years ago engineering students learning electronics could have had more solid hydraulics background knowledge, but I bet these days more people start from electronics before doing hydraulics.

You are correct. The best way I've found to build electrical intuition is to build circuits and experiment with them. It takes time.

I learned electronics as a kid from a kit that used a water analogy to describe things. The transistor was described as a vane across a large water canal, connected with a hinge to a smaller vane in a smaller water canal, which allowed it to move up and down. So, a small water current in the small canal could cause a large water current in th large canal. Did it fully and accurately describe the gain characteristics of a real transistor ? No, it didnt. But it would be ridiculous to dismiss it entirely as a learning tool because of it. Overall I think the hydraulic analogy is a very good one, it holds up well and it breaks down only in the face of quantum phenomenon. It allows to understand electricity using concepts children are familiar with. All children like to play with water and have some intuitive sense of it. By the time you care about quantum electronics you can also understand the limitations of the analogy.

Maybe some people just aren't smart enough to understand quantum tunnelling and band gaps, and there's no real better analogy. And even if you do kind of get these, the bad analogies let you remember the standard behaviour so you can reason about what the micro details are.

The problem is in my opinion not the hydraulic analogy itself, but the fact that it is usually presented as a model.

It has its place, but it should come with a big disclaimer at the front that it is not a model where we assume it is incomplete and which does not explain everything but an analogy that is fundamentally wrong in everything and only meant to help us understand the one particular aspect of the real thing.

But it IS a model. It says that the model you use for water is the same as the one for electricity.

I actually like it because it has no pretense that you are actually doing/learning science. Compare that to the simplified models in high school physics where you have the government literally lying to children about how physical phenomena work (e.g. Newton's cradle) and nobody cares despite the teacher's insistence that you "need to know" this misinfo.

If you want to complain about that model for water being wrong though I'm all ears. But the model for water also being the model for electricity is just obviously wrong, cause they aren't the same thing, if both were correct then there would only be a single word.

"all models are wrong, some are useful."

-George Box

Analogies aren’t perfect, but they’re still useful. How else are you gonna explain to a grade schooler the difference between voltage and current? This article suggests no alternative.

This.

The premise of the article is that the hydraulic analogy breaks down for semiconductors.

My reaction: no kidding, a classical explanation doesn’t cover fundamentally quantum devices.

The whole premise that your model should only need small tweaks to evolve seems wrong. The author has probably forgotten their first encounter with quantum mechanics. It’s not a small tweak. It’s a bewildering change.

Is it particularly wrong to be using hydraulics to explain MOSFETs? I’m not qualified to have an opinion on this; maybe someone is holding on to the analogy too long. For classical effects it works well.

Obviously you bust out your university physics book and work through the equations with them! There is nothing my 9 year old loves more than doing derivatives!

We didn’t even teach her the equations for velocity or distance, we derived them starting from Jerk and integrated our way up the food chain. And none of that linear single dimension nonsense either, kiddo learned in both 3D and polar coordinates from day one; with friction too because like spherical cows, friction-free is a lie.

Teaching grade schoolers single-dimension, frictionless physics really messes them up once they get into grad school and start doing real math.

Counterpoint: Youtuber AlphaPhoenix does a very good job explaining electricity using the hydraulic analogy https://www.youtube.com/watch?v=X_crwFuPht4

Related video by Steve Mould about "Spintronics" [1]

    Mechanical circuits: electronics without electricity

https://youtu.be/QrkiJZKJfpY

[1]: https://upperstory.com/spintronics/

Well, quite a few things can be described using the same mathematics - including switches, resistors, capacitors, and inductive coils. Sure, it is not everything, but well - fluid is fluid, electricity is electricity.

For a small simulation, see http://sjbyrnes.com/1235/.

The problem with the hydraulic analogy is that people take it too far.

It’s super useful when describing simple ideas to children or laypeople that only are ever going to look at passive components in dc circuits, or at systems so simple it is unnecessary to distinguish between ac and dc.

It quickly falls apart as you involve more complex concepts.

But it is very useful at the basic, basic level for developing simple intuitions about voltage, amperage, and resistance. As long as it is used in-scope and with the appropriate caveats, it is very helpful.

Taking it too far is the problem. You can’t use Newtonian physics to build a GPS system, and you shouldn’t try to go beyond basic eli5 electricity with the hydraulic model.

The water flow analogies always messed me up because young literal me couldn’t handle reconciling the abstraction with “actually current flows negative to positive”

The hydraulic analogy works great for explaining the basics of electricity to regular people. My stepson doesn't know a transistor from the Trans-Siberian Orchestra, but he needs to know how to use a multimeter to work on his motorcycle.

All models are wrong, but some are useful.

I thought you just used hydro on simple RLC circuits, and kept it at that?

For semiconductors there's no reason to invent a hydro equivalent that doesn't exist, what's the point? You want a model to be intuitive to the student, and a made-up thing won't be.

As you go further with EE, things also tend to turn into a sort of logic gate puzzle rather than a continuous (voltage-current) problem. That's a while other analogy whose limits you also need to know.

say no to manipulated scrollbars

Reply on news.ycombinator.com