What makes software unique to other engineering disciplines is that it isn't a discipline at all. What makes software so great is how quick the iteration cycles are.
Software sits at a higher abstraction level than physical hardware, so much of our time is spent throwing at the wall and seeing what sticks because that's often (although not always) the best use of time.
These days you are hard pressed to find a language that uses dynamic scoping as the default.
The needle likes to swing back and forth between dynamic be static typing but we definitely seem to be coming back to static typing again
It's also my experience that a few well written integration tests are more useful and catch more actual bugs than a million unit tests. This is especially true of web apps which are inherently highly stateful.
The most powerful design paradigm I have ever found is to spend the first, say, 10-20% of my time making a rough prototype. Not with an eye of actually using a single line of it (although it happens) but just to do the quickest and dirtiest possible exploration of what the actual problem is.
you mean concerns that should be seperated but they did it in a wrong way?
patients = 400
x = pat / 2
And x would be 200. This seems like an obvious footgun today and I don't think any language designer would even consider it, but it seemed to make sense for a while.This joke is like a decade old at this point https://xkcd.com/1172/
Namely, what does the following code do?
patients = 400
patents = 20
x = patients / 10
If the compiler truncates, x will be 2 (assuming it allows variable reassignment), and you've probably introduced at least two bugs. It's doing some sort of lookup against the source, well, all bets are really off.“Doing something” is the beginning of how to describe the actual truth of what some actions causes in relation to its affect on reality.
“Supporting something” has literally nothing to do with reality beyond what’s necessary to support a brain that can _intend_ a result or action regardless of what actually happens.
I linked the xkcd comic because the behavior described by the developer in the comic met the criteria for “doing something” but it didn’t meet the criteria for “supporting something” because the behavior of the software was not what the developer intended.
Back in my EE days in 90's, "full stack engineer" meant, for example, being able to build a physical calculator by connecting a numerical keyboard, bunch of 7 segment displays and a micro controller using bunch of wires on a circuit board, and THEN writing the assembly code to allocate memory and run your program in an infinite loop. You had to erase your EPROM with UV and burn your program to the chip over and over every time you changed a byte in your code. Debugging? You wish.
As a side effect, full stack engineers had the risk of getting electrocuted, or going blind :)
Which, indeed, is different from engineering where constraints are non-negotiable, and thus the methodology as well.
I think a lot of people doing functional programming, as an example, enjoy the constraints and the discipline that it imbues on their craft.
Then marketing come along and ask to make it a trilogy and a few weeks before release they ask you to add in Ewoks because it'll sell more toys.
except that's not really true. The big difference is that to negotiating the constraints when it comes to (physical) engineering, one of the biggest factors is money (and lots of it). "Well, we could double the span but for that you'll need to have parts A1, D5 and T3 built out of ..."
Software development isn't free, but it doesn't typically incur doubling of costs for tweaks to the feature list (and I do mean tweaks).
The _need_ to build higher-level abstractions to manage the complexity or verboseness required to do things from lower level components always seems to be fighting with the _desire_ to build your own higher-level abstractions that fit your own view of how things 'should be', despite many of the decisions going into this being abstract and not easily objectively measurable.
On top of that, software components as 'reusable boxes' only seems to work up to a certain level. The idea seems to be that higher-level abstractions and reusable pieces are all nicely shaped square boxes that all perfectly fit inside other boxes, but the reality seems to be more that even the best reusable boxes aren't perfectly square, have weird edges, and themselves have to fit into larger non-square shapes with weird edges.
And what are the weird sharp edges? Decisions that had to be made about solving a problem. Every higher-level software component is on some level a set of (mostly arbitrary, sometimes measured) decisions about how to take a set of lower level generic components and make something more specific with them. Libraries might assume a certain structure but leave choices for how they're composed to the user. Applications make many more decisions about how their components are used in order to provide a more specific tool. There is a move from genericity to specificity as you move up layers of components.
And finally, software requirements change all the time, but the need to change is itself at odds with the requirement that problems are solved only by making hundreds of decisions that have to work together for a working solution but also may be difficult to undo.
> the constraints and the discipline that it imbues on their craft.
Perhaps the constraints and discipline here comes from a desire to move towards some kind of standard or expected way of structuring software, knowing from experience that the cognitive cost of making (and sometimes having to read and understand) all the arbitrary decisions mentioned above is actually quite high. (This itself also leads to the notion of 'best practices' which seemingly change too often to ever really be considered best).
Different jurisdictions have different tax rates.
Where did I say that? There are definitely people who straight up break the relevant laws and don't pay their taxes.
Of course, whether someone breaks a law or not, is an issue independent of 'fairness'. But I don't know what you mean by 'fair'.
However, at the end of the day, there is an input and output and compute and memory needed to run the thing and if we look at that we realize, we never actually left the bounded physical realm and we can still engineer software systems against real world constraints. We can judge its efficiency and breaking points.
What's very different is the cost to change the system to do something new and that's where this unbounded complexity blows up in our face.
This is a common sense view of computation that's unfortunately wrong.
The simplest counter example is the busy beaver program: with as little as 12 states we have saturated the computational capabilities of the universe, but it looks completely safe and sane for the first few states you would be testing against.
You may call it pathological, and you'd be right, but the point is that you never know under which rug a function that takes more computation than the universe can supply is hiding.
By comparison power electronics engineers don't have to formally prove that they didn't accidentally include a nuclear power plant in their e-scooter design.
If on the other hand you're speaking of the theoretical computational needs of the program you just wrote, then your earlier dismissal of mathematics and its "even worse track record" is all the sillier.
If I was designing a software system, I could introduce a time constraint. An imagined conversation: "How long will it take to get an answer? Between half a second and the heat death of the universe. OK. Can we just issue a timeout error after 1 second?"
This is putting controls in place so the system doesn't exceed its constraints and although hypothetical it might be able to do a job for any input, it can't because we haven't been able to find a more efficient solution for certain known and unknown scenarios.
Let's take a made up example of a structural engineer designing a building. As another comment mentioned, in theory the design space here is enormous. Just the concrete mix can be endlessly optimized. But here there's a large monetary cost tradeoff that's obvious to everyone involved. Say an overenthusiastic junior proposes attempting to rediscover ancient Roman concrete mix for this project. Everyone from the other engineers to management can call that out as absurd.
In theory you can make the building out of anything. But in practice the economics of producing real world components only allows for a few choices in each stage of the design.
Meanwhile, software components are essentially free, especially if they're open source. Just clone the repo, hook up the code and you're done, right? So surprisingly often, the overenthusiastic junior can convince the whole company to build the new feature with some obscure framework that'll be unsupported in a few years. And no one can reliably call them out, because there's no easy, objective way to measure something like tech debt. (Another compounding problem is the rapid growth of the software field. Juniors are minted faster than they can be trained.)
To make the software example more concrete: there's an internal config language "blub" in one of the FAANGs that someone designed in two months. It looked simple, gained a lot of adoption quickly. Fast forward 5 years, and it turns out using blub was a big mistake. It scales terribly, the semantics were badly thought out and cause subtle bugs everywhere. But here's the kicker: there are now millions of lines of blub. Over the next decade, the company makes several very expensive attempts to replace blub and fails. Blub continues to underpin their production infra to this day.
So to sum up, I think maybe the core issue is not so much "3d space versus hyperbolic space". If you view the space as the tree of decisions you need to make to reach the goal, all engineers are working in hyperbolic space.
However, when designing real world things, economics rapidly culls the design tree. Buying materials requires cash up front. But in software it's often the opposite: you get the libraries "for free", and the unmeasurable tech debt accrues over time.
Exactly like sticking with blub, which was only chosen because of historical accident, but they're stuck with it because it would cost too much to stop all progress while refactoring everything to work around it.
You're arguing a matter of degree while asserting a matter of kind.
And the point is still ridiculous with the explicit comparison to maths, which is somehow different because there exists wrong proofs.
Trying to diminish engineering disciplines that have existed for millennia by imposing an arbitrary constraint of "3 euclidean" is pretty wank. Let's drop silly constraints.
Do you have any idea how complicated concrete is? It's just sand, aggregate, cement and water. Exothermic reaction. Job done.
Let's look at wood ... the stuff is a bundle of fibres and stuff and yet we build huge structures out of it. Who on earth knows how or why plywood works? Britain had several all wooden aircraft during WWII - the Mosquito was so good that Herman Goering declared that it was worth two kills.
Steel - its just iron with other stuff.
Software engineering is growing up gradually but please don't be a dick towards people who practice disciplines that invented things and concepts you rely on and live within.
For example the concept of a token that allows you to do something exclusively - " exclusive lock" - was popularised by early railways. A single track should have only one train on it and after a few accidents the idea of an exclusive token was "invented" that could be passed across to the next exclusive user.
When you say hyperbolic, you should be aware of how close that is to hyperbole.
The point is that all other engineering disciplines don't have enough design space to make a true mess of their projects. Software by comparison hasn't been limited in its design space since the first super computers started having gigabytes of ram memory in the late 1980s.
That doesn't mean that either discipline is better or worse, it means they are different. Other than math there is no other field which has as much freedom when it comes to basic design choices as software engineering, and math proofs are even more of a mess than software programs, to the point that pretty much every non-formally verified proof is wrong in some way.
Or to put another way: would railways be more or less complex if we allowed them to be embedded in a higher dimensional space instead of being essentially a 2d network on the earth surface? If you don't think so I'd like you to think about what a train schedule would look like on hyper graphs: https://en.wikipedia.org/wiki/Hypergraph
Every discipline has infinite depth.
Railways in 2D may seem simple but there the simplicity may be intentional. Just like when you write software that runs things like planes, trains, or medical devices, all of a sudden you’re writing in C, without recursion, without function pointers, without using the heap, etc.
Look at the concord airplane. A mess of a project.
Look at anything in biology and human health. The complexity is unimaginable compared to software.
Edit: reduced emotions
Since I want programming to be inclusive and not have a bad rap publicly, when a peer says “software is better than all other disciplines because of X, Y, Z” I want to convince them otherwise.
As a leader and teacher with more grey hair every day, I feel this responsibility
You should probably do less of that since:
1). That's not what any of my posts said.
2). My degree says mathematical physics and not computer science or software engineering.
3). A field where you're tone policed by the tone deaf is the opposite of welcoming and inclusive.
U/callc appears to be reacting to the public perception of how this thread would read externally, you appear to be reacting based on hard logic, which is unpalatable to most of the public.
I think you two are talking past each other
This is why steelmanning a comment before arguing against it should be a rule.
Based on your comment I’m interpreting “special” as “better”, or “harder”.
I interpreted the comment you replying to as software is “special” as in it is “different” and the differences require different solutions than other fields have.
Like when I talk to my civil engineer friends I only on the surface comprehend what they are dealing with when they need to move literal thousands of tons of mass. I can try and make an analogy to my own experience with bandwidth and memory, but nothing will be a 1:1 comparison with the amount of energy used when the physical mass of the entire internet is within an order of magnitude of a strawberry.
Simultaneously my civil engineers friends don’t have to deal with a situation of one of their apprentices making a logical mistake based on a single missed `!` and their features suddenly using up 100% of all resources everywhere
Any type of physical engineering is based on hard facts, data, and well established historical research, mathematics, and more.
I've seen many an electrical engineer say, "fuck it, this is too hard and pays too little" and so they pivoted into software engineering quite easily.
On the other hand I've never seen the opposite. Real engineering is hard because you actually have to get shit perfect. The wrong o rings and real people die.
There are a few real software engineering shops and real software engineers, but 90% are just slinging shit together for a bunch of enterprise CRUD and they are all just one bad deploy away from disaster.
I read it as: “software is special because its design space is high dimensional, compared to all other engineering disciplines which are limited to 3 spatial dimensions. Large design space leads to ability to mess things up.”
I wholeheartedly agree with you: “real” engineering (where lives are on the line) is hard. Most software that is made is not to control pace makers, air plane autopilot, space shuttles, etc. When we do need to program in these life and death situations, we should look carefully at the real engineers that do this all the time (looking at you self-driving car software).
Engineering with constraints builds discipline. Maybe we are lacking as engineers in software because the constraint bar just continues to raise.
We could always talk about stacks, AVL trees, or NLP before any meaningful conception of computers, and indeed a lot of computer science research happened when computers were still fairly primitive. Computer science and its applied form, software engineering, are about computation (data and code) and abstraction. Even computation is just a distraction at some level. There's some world with the ideal behavior that a program should abstract from, but an ideal is an ideal, so programming is often floundering to figure out what ad-hoc world the program mirrors and tweaking haphazardly. Software engineering isn't engineering not because it can't be, due to the (very real) differences of software from the physical world. It isn't engineering currently because we aren't taking control of our actions and their consequences, in all senses.
Intellectual superiority is not relevant here. Everyone has hard tasks. If you're a civil engineer, then design and build infrastructure well. If you're a programmer, then design and build programs well. Any discipline demands honesty and forthrightness in evaluating its work.
I totally agree with their post and get what they're saying.
Within circuits (big or small) it is possible to do some crazy stupid shit, but reality kicks in much sooner. Either because of physics or price.
Analog circuits (my specialty) are a perfect example of this. Sure I can try to cut corners and interconnect two distant parts of the circuit together in weird ways ("tight coupling" in software), but the options I will reach for are limited, and "nature" forces me to "decouple" them. With software I can just set a distant variable...
I have a Building Engineer on my team, who also moved to Software Engineering. I can ask for examples in his field too.
90% of programming is enterprise CRUD, and the bar is low. Likewise, 90% of electrical work is basic circuitry with off the shelf parts: does it work and not catch fire. There are electrical engineers doing Real :TM: Engineering, but 90% or more of the work is electrician work.
Sorry, what? This is an extraordinary claim.
If you do not have the experience and skill to express your ideas in text in a clear and concise, you will struggle to write good code (and vice versa tbf).And coders generally are lacking on the more humanities side of their educations.
Case in point - more than one Mars mission has failed due to software errors.
However you could argue that the best way to deal with that complexity is not to have a brilliant mind that can grok more of that complex space, but by simply taking good engineering practices to minimise and manage it - ultimately the complexity is beyond us all if not managed.
Software is ultimately limited to building things in a space that is only countably infinite. There is only so much junk you can hide in a space that is only countably infinite.
All other engineering disciplines by comparison deal in a space that is uncountably infinite and you can hide _anything_ in such a space without it being obvious. This is exemplified by the unpleasant discovery all of us have had of a supposedly measurable "collection" of intervals having such weird holes all over the real number line and the near impossibility of shifting it up or down that makes sense for it without having to rethink our whole notion of measure and still end up with something like the counting measure that definitely doesn't make sense.
People, including myself, have a seriously bad intuition just how much volume there is in an uncountably infinite space.
I think that you have a basic grasp of the word "dimension" but use the word "euclidean" without understanding what it means just because you heard it in conjunction with "dimension".
If you think that e.g. building a bridge is a three dimensional problem in the sense that you are actually talking about (which has nothing to do with euclidean geometry), you lack the basic understanding of what you are talking about from which you could draw any conclusion like that with any sense of intellectual honesty.
All other engineering disciplines are ultimately limited to building things in (at most) 3 euclidean dimensions. There is only so much junk you can hide in a finite volume of space.
Code by comparison lives in hyperbolic space [0] and you can hide _anything_ in such a space without it being obvious. This is exemplified by the unpleasant discovery all of us have had of a supposedly peripheral folder holding source code called all over the code base and the near impossibility of moving it in a location that makes sense for it without having to refactor the whole code base.
People, including myself, have a seriously bad intuition just how much volume there is in a space which grows at least exponentially.
The closest discipline to software engineering is mathematics and that has an even worse track record. There's the folklore about half of all math papers giving the wrong proof for the right conclusion. By comparison software engineering only gets catastrophic bugs less than every other time a program is run.
[0] All trees are natively embedded in some hyperbolic space of whatever curvature matches the average number of children per node, and all code can be ultimately represented as a tree.