FRESH
Hacker News
Hacker News Hacker NewsThere are few principle of software engineering that I hate more than this one, though SOLID is close.
It is important to understand that it is from a 1974 paper, computing was very different back then, and so was the idea of optimization. Back then, optimizing meant writing assembly code and counting cycles. It is still done today in very specific applications, but today, performance is mostly about architectural choices, and it has to be given consideration right from the start. In 1974, these architectural choices weren't choices, the hardware didn't let you do it differently.
Focusing on the "critical 3%" (which imply profiling) is still good advice, but it will mostly help you fix "performance bugs", like an accidentally quadratic algorithms, stuff that is done in loop but doesn't need to be, etc... But once you have dealt with this problem, that's when you notice that you spend 90% of the time in abstractions and it is too late to change it now, so you add caching, parallelism, etc... making your code more complicated and still slower than if you thought about performance at the start.
Today, late optimization is just as bad as premature optimization, if not more so.
“A variable should mean one thing, and one thing only. It should not mean one thing in one circumstance, and carry a different value from a different domain some other time. It should not mean two things at once. It must not be both a floor polish and a dessert topping. It should mean One Thing, and should mean it all of the time.”
- Every website will be vibecoded using Claude Opus
This will result in the following:
- The background color will be a shade of cream, to properly represent Anthropic
- There will be excessive use of different fonts and weights on the same page, as if a freshman design student who just learned about typography
- There will be an excess of cards in different styles, a noteworthy amount of which has a colored, round border either on hover or by default on exactly one side of the card
"In analyzing complexity, fast iteration almost always produces better results than in-depth analysis."
Boyd invented the OODA loop.
> "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it"
Structure code so that in an ideal case, removing a functionality should be as simple as deleting a directory or file.
"Every application has an inherent amount of irreducible complexity that can only be shifted, not eliminated."
I think it's more useful to remember when you're refactoring that if you try to make one bit of a system simpler then you often just make another part more complex. Why write something twice to end up with it being just as bad the other way round?
Asking "who wrote this stupid code?" will retroactively travel back in time and cause it to have been you.
"The meta-law of software engineering: All laws of software engineering will be immediately misinterpreted and mindlessly applied in a way that would horrify their originators. Now that we can observe the behaviour of LLMs that are missing key context, we can understand why."
Or, you can't boil down decades of wisdom and experience into a pithy, 1 sentence quote.
> Fen's law: copy-paste is free; abstractions are expensive.
edit: I should add, this is aimed at situations like when you need a new function that's very similar to one you already have, and juniors often assume it's bad to copy-paste so they add a parameter to the existing function so it abstracts both cases. And my point is: wait, consider the cost of the abstraction, are the two use cases likely to diverge later, do they have the same business owner, etc.
- Shirky Principle: Institutions will try to preserve the problem to which they are the solution
- Chesterton's Fence: Changes should not be made until the reasoning behind the current state of affairs is understood
- Rule of Three: Refactoring given only two instances of similar code risks selecting a poor abstraction that becomes harder to maintain than the initial duplication
When it comes to frameworks (any framework) any jargon not explicitly pointing to numbers always eventually reduces down to some highly personalized interpretation of easy.
It is more impactful than it sounds because it implicitly points to the distinction of ultimate goal: the selfish developer or the product they are developing. It is also important to point out that before software frameworks were a thing the term framework just identifies a defined set of overlapping abstract business principles to achieve a desired state. Software frameworks, on the other hand, provide a library to determine a design convention rather than the desired operating state.
This one is listed as design, but it could just as easily count as architecture. Guessing a lot developers have worked on scaling with lambda functions or a complex IAC setup when a simple API running on a small VPS would have done the trick, at least until enough people are using the application for it to be considered profitable.
People bring it up to argue for never thinking about performance, which flips the intent on its head. The real takeaway is that you need to spot that critical 3% early enough to build around it, and that means doing some optimization thinking up front, not none at all.
9. Most software will get at most one major rewrite in its lifetime.
I actually had a college run over by a bus on the way to work in London, was very lucky and made a full recovery.
Head poking out under the main exit of the bus.
(~150) is the size of a community in which everyone knows each other’s identities and roles.
In anthropology class. You can ask someone to write down the name of everyone they can think of, real or fictional, live or dead and most people will not make it to 250.
Some individuals like professional gossip columnists or some politicians can remember as many as 1,000 people.
Then I committed the code and let the second AI review it. It too had no problem with goto's.
Claude's Law: The code that is written by the agent is the most correct way to write it.
Or develop a skill to make it correct, fast and pretty in one or two approaches.
complexity(system) =
sum(complexity(component) * time_spent_working_in(component)
for component in system).
Separately, I'd add Rust's API design principles, though it's more of a adjunct with things in common. https://gist.github.com/mjball/9cd028ac793ae8b351df1379f1e72...
Where's Chesterton's Fence?
https://en.wiktionary.org/wiki/Chesterton%27s_fence
[EDIT: Ninja'd a couple of times. +1 for Shirky's principle]
Sure, don't add hooks for things you don't immediately need. But if you are reasonably sure a feature is going to be required at some point, it doesn't hurt to organize and structure your code in a way that makes those hooks easy to add later on.
Worst case scenario, you are wrong and have to refactor significantly to accommodate some other feature you didn't envision. But odds are you have to do that anyway if you abide by YAGNI as dogma.
The amount of times I've heard YAGNI as reasoning to not modularize code is insane. There needs to be a law that well-intentioned developers will constantly misuse and misunderstand the ideas behind these heuristics in surprising ways.
A couple are well-described/covered in books, e.g., Tesler's Law (Conservation of Complexity) is at the core of _A Philosophy of Software Design_ by John Ousterhout
https://www.goodreads.com/en/book/show/39996759-a-philosophy...
(and of course Brook's Law is from _The Mythical Man Month_)
Curious if folks have recommendations for books which are not as well-known which cover these, other than the _Laws of Software Engineering_ book which the site is an advertisement for.....
I wish AWS/Azure had this functionality.
I wonder if it should be called "Law of Leaky Metaphors" instead. Metaphor is not the same thing as Abstraction. I can understand a "leaky metaphor" as something that does not quite make it, at least not in all aspects. But what would be a good EXAMPLE of a Leaky Abstraction?
Most of them are also wrong
Especially things like “every system grows more complex over time” — you can see it in almost any project after a few iterations.
I think the real challenge isn’t knowing these laws, but designing systems that remain usable despite them.
> When describing phenomena in the social world
> Software Engineers gravitate towards eponymous 'laws'.
* https://martynassubonis.substack.com/p/5-empirical-laws-of-s...
* https://newsletter.manager.dev/p/the-unwritten-laws-of-softw..., which linked to:
* https://newsletter.manager.dev/p/the-13-software-engineering...
While browsing it, I of course found one that I disagree with:
Testing Pyramid: https://lawsofsoftwareengineering.com/laws/testing-pyramid/
I think this is backwards.
Another commenter WillAdams has mentioned A Philosophy of Software Design (which should really be called A Set of Heuristics for Software Design) and one of the key concepts there are small (general) interfaces and deep implementations.
A similar heuristic also comes up in Elements of Clojure (Zachary Tellman) as well, where he talks about "principled components and adaptive systems".
The general idea: You should greatly care about the interfaces, where your stuff connects together and is used by others. The leverage of a component is inversely proportional to the size of that interface and proportional to the size of its implementation.
I think the way that connects to testing is that architecturally granular tests (down the stack) is a bit like pouring molasses into the implementation, rather than focusing on what actually matters, which is what users care about: the interface.
Now of course we as developers are the users of our own code, and we produce building blocks that we then use to compose entire programs. Having example tests for those building blocks is convenient and necessary to some degree.
However, what I want to push back on is the implied idea of having to hack apart or keep apart pieces so we can test them with small tests (per method, function etc.) instead of taking the time to figure out what the surface areas should be and then testing those.
If you need hyper granular tests while you're assembling pieces, then write them (or better: use a REPL if you can), but you don't need to keep them around once your code comes together and you start to design contracts and surface areas that can be used by you or others.
My bet is on the long arc of the universe trending toward complexity... but in spite of all this, I don't think all this complexity arises from a simple set of rules, and I don't think Gall's law holds true. The further we look at the rule-set for the universe, the less it appears to be reducible to three or four predictable mechanics.
- Not realizing it's a very concrete theorem applicable in a very narrow theoretical situation, and that its value lies not in the statement itself but in the way of thinking that goes into the proof.
- Stating it as "pick any two". You cannot pick CA. Under the conditions of the CAP theorem it is immediately obvious that CA implies you have exactly one node. And guess what, then you have P too, because there's no way to partition a single node.
A much more usable statement (which is not a theorem but a rule of thumb) is: there is often a tradeoff between consistency and availability.
The UX pyramid but applied to DX.
It basically states that you should not focus in making something significant enjoyable or convenient if you don't have something that is usable, reliable or remotely functional.
Because rewriting old complex code is way more time consuming that you think it'll be. You have to add not only in the same features, but all the corner cases that your system ran into in the past.
Have seen this myself. A large team spent an entire year of wasted effort on a clean rewrite of an key system (shopping cart at a high-volume website) that never worked... ...although, in the age of AI, wonder if a rewrite would be easier than in the past. Still, guessing even then, it'd be better if the AI refactored it first as a basis for reworking the code, as opposed to the AI doing a clean rewrite of code from the start.
As JFK never said:
“””We do these things, not because they are easy,
But because we thought they would be easy”””
ha, someone needs to email Netlify...
https://web.archive.org/web/20260421113202/https://lawsofsof...
> The first 90% of the code accounts for the first 90% of development time; the remaining 10% accounts for the other 90%.
It should be 90% code - 10% time / 10% code - 90% time
Just because some things were observed frequently during a certain period, doesn't mean it's a "Law" or even a "Principle"; it's merely a trend.
Relax. You will make all the mistakes because the laws don't make sense until you trip over them :)
Comment your code? Yep. Helped me ten years later working on the same codebase.
You can't read a book about best practises and then apply them as if wisdom is something you can be told :)
It is like telling kids, "If you do this you will hurt yourself" YMMV but it won't :)
YAGNI and "you will ship the org chart" are the two most commonly useful things to remember, but they aren't laws.
> Any sufficiently complicated C or Fortran program contains an ad hoc, informally-specified, bug-ridden, slow implementation of half of Common Lisp.
This one belongs to history books, not to the list of contemporary best practices.
> The less you know about something, the more confident you tend to be.
From the first line on the wiki article:
> systematic tendency of people with low ability in a specific area to give overly positive assessments of this ability.
Or, said another way, the more you know about something the more complexities you're aware of and the better assessment you can make about topics involving such. At least, that's how I understand it in a nutshell without explaining the experiments run and the observations that led to the findings.
> Premature Optimization (Knuth's Optimization Principle)
> Another example is prematurely choosing a complex data structure for theoretical efficiency (say, a custom tree for log(N) lookups) when the simpler approach (like a linear search) would have been acceptable for the data sizes involved.
This example is the exact example I'd choose where people wrongly and almost obstinately apply the "premature optimization" principles.
I'm not saying that you should write a custom hash table whenever you need to search. However, I am saying that there's a 99% chance your language has an inbuilt and standard datastructure in it's standard library for doing hash table lookups.
The code to use that datastructure vs using an array is nearly identical and not the least bit hard to read or understand.
And the reason you should just do the optimization is because when I've had to fix performance problems, it's almost always been because people put in nested linear searches turning what could have been O(n) into O(n^3).
But further, when Knuth was talking about actual premature optimization, he was not talking about algorithmic complexity. In fact, that would have been exactly the sort of thing he wrapped into "good design".
When knuth wrote about not doing premature optimizations, he was living in an era where compilers were incredibly dumb. A premature optimization would be, for example, hand unrolling a loop to avoid a branch instruction. Or hand inlining functions to avoid method call overhead. That does make code more nasty and harder to deal with. That is to say, the specific optimizations knuth was talking about are the optimizations compilers today do by default.
I really hate that people have taken this to mean "Never consider algorithmic complexity". It's a big reason so much software is so slow and kludgy.
Here's another law: the law of Vibe Engineering. Whatever you feel like, as long as you vibe with it, is software engineering.
That one's free.
In most places, people don't follow this rule, as it ensures either you're working an extra 10-20 hours a week to keep things clean, or stuck at mid-level for not making enough impact.
I choose the second option. But I see people who utterly trash the codebase get ahead.
Site not available This site was paused as it reached its usage limits. Please contact the site owner for more information.
- NIH
- GIGO
- Rule of 3
https://lawsofsoftwareengineering.com/laws/premature-optimiz...
It leaves out this part from Knuth:
>The improvement in speed from Example 2 to Example 2a is only about 12%, and many people would pronounce that insignificant. The conventional wisdom shared by many of today’s software engineers calls for ignoring efficiency in the small; but I believe this is simply an overreaction to the abuses they see being practiced by penny-wise- and-pound-foolish programmers, who can’t debug or maintain their “optimized” programs. In established engineering disciplines a 12% improvement, easily obtained, is never considered marginal; and I believe the same viewpoint should prevail in software engineering. Of course I wouldn’t bother making such optimizations on a one-shot job, but when it’s a question of preparing quality programs, I don’t want to restrict myself to tools that deny me such efficiencies.
Knuth thought an easy 12% was worth it, but most people who quote him would scoff at such efforts.
Moreover:
>Knuth’s Optimization Principle captures a fundamental trade-off in software engineering: performance improvements often increase complexity. Applying that trade-off before understanding where performance actually matters leads to unreadable systems.
I suppose there is a fundamental tradeoff somewhere, but that doesn't mean you're actually at the Pareto frontier, or anywhere close to it. In many cases, simpler code is faster, and fast code makes for simpler systems.
For example, you might write a slow program, so you buy a bunch more machines and scale horizontally. Now you have distributed systems problems, cache problems, lots more orchestration complexity. If you'd written it to be fast to begin with, you could have done it all on one box and had a much simpler architecture.
Most times I hear people say the "premature optimization" quote, it's just a thought-terminating cliche.
"Before SpaceX, launching rockets was costly because industry practice used expensive materials and discarded rockets after one use. Elon Musk applied first-principles thinking: What is a rocket made of? Mainly aluminum, titanium, copper, and carbon fiber. Raw material costs were a fraction of finished rocket prices. From that insight, SpaceX decided to build rockets from scratch and make them reusable."
Everything including humans are made of cheap materials but that doesn't convey the value. The AI got close to the answer with it's first sentence (re-usability) but it clearly missed the mark.
Law 0: Fix infra.