One use case that I'm beginning to find useful is to go into a specific directory of code that I have written and am working on, and ask the AI agent (Claude Code in my case) "Please find and list possible bugs in the code in this directory."
Then, I can reason through the AI agent's responses and decide what if anything I need to do about them.
I just did this for one project so far, but got surprisingly useful results.
It turns out that the possible bugs identified by the AI tool were not bugs based on the larger context of the code as it exists right now. For example, it found a function that returns a pointer, and it may return NULL. Call sites were not checking for a NULL return value. The code in its current state could never in fact return a NULL value. However, future-proofing this code, it would be good practice to check for this case in the call sites.
This all reminds me of one of the most foundational and profound papers ever written about software development: Peter Naur's "Programming as Theory Building". I have seen colleagues get excited about using Claude to write their software for them, and then end up spending at least as much time as if they had written it themselves trying to develop a theory the code that was produced, and an understanding sufficient to correct the problems and bugs in the created code.
Every professional software engineer confronts the situation of digging into and dealing with a big wad of legacy code. However, most of us prefer those occasions when we can write some code fresh, and develop a theory and deep understanding from the get-go. Reverse-engineering out a sufficient theory of legacy code to be able to responsibly modify it is hard and at times unsatisfying. I don't relish the prospect of having that be the sum total of all my effort as a software engineer, when the "legacy code" I need to struggle to understand is code generated by an AI tool.
The original comment by Rob Pike and discussion here have implied or used the word "evil".
What is a workable definition of "evil"?
How about this:
Intentionally and knowingly destroying the lives of other people for no other purpose than furthering one's own goals, such as accumulating wealth, fame, power, or security.
There are people in the tech space, specifically in the current round of AI deployment and hype, who fit this definition unfortunately and disturbingly well.
Another much darker sort of of evil could arise from a combination of depression or severe mental illness and monstrously huge narcissism. A person who is suffering profoundly might conclude that life is not worth the pain and the best alternative is to end it. They might further reason that human existence as a whole is an unending source of misery, and the "kindest" thing to do would be to extinguish humanity as a whole.
Some advocates of AI as "the next phase of evolution" seem to come close to this view or advocate it outright.
To such people it must be said plainly and forcefully:
You have NO RIGHT to make these kinds of decisions for other human beings.
Evolution and culture have created and configured many kinds of human brains, and many different experiences of human consciousness.
It is the height (or depth) of arrogance to project your own tortured mental experience onto other human beings and arrogate to yourself the prerogative to decide on their behalf whether their lives are worth living.
(Show HN?)
Here is another attempt at providing intuition on complex numbers, the multiplication rule in particular:
"Why does complex multiplication have to be the way it is?"
https://gregfjohnson.com/complex/
I find that the easiest intuitive on-ramp to complex arithmetic is to start with compass headings: "Oh that nice coffee shop? Go two blocks north and then a block east." Numbers come with any direction on the compass, not just "east" and "west". It turns out that it is pretty easy to intuitively justify multiplication by a scalar and addition of complex numbers, but multiplication is harder. A great way to get a feel for multiplication is to consider the equation "(x+1)(x-1) = x*2 - 1". Then, substitute "i" for x. The left-hand side is (intuitively) on a circle of radius 2 centered at the origin, and the right-hand side is on a circle of radius 1, where the circle is shifted horizontally so that its center is on the real line at -1. There's only one place these two circles meet: -2 on the real line.
There are bitter and heartstrong arguments on this topic. Brachycephalic breeds with flat faces or short snouts (for example pugs, bulldogs, boxers, Boston terriers) cause anger among some and devotion among others. (We are a three-pug family.) Pugs have an average lifespan of 12 to 15 years, as compared with the overall average dog lifespan of 10 to 13 years. We have also had (and loved) boxers, which have an average lifespan of 10 to 12 years. Boxers seem to be genetically prone to cancer, which is how we lost our most recent boxer.
There's a classic cartoon showing two wolves in the bushes at the edge of a campfire, looking at the leftovers being thrown around by the humans. One says, "Look, what the heck, let's cozy up to these two-legged creatures that seem to have lots of food. What could go wrong?"
Next frame is a picture of an unhappy-looking pug wearing a birthday hat..
I think the reason take so much offense at e.g. pugs is because they don't need to be bred that way. Where I'm at a law was passed making it illegal to breed dogs with congenital conditions. As a result basically all the pugs you see here have a bit of a snout as do all the boxers. Crucially this didn't change the breeds personality perceptibly, and it only changes the look the tiniest of bit. In trade you get dogs that are visibly happier and have more energy.
A neighbor in particular previously had two french bulldogs with no snout. They'd spend all day panting and snoring. Once they passed away they got new ones, now with snout. They spend all day running and jumping instead.
If I'm reading your meaning correctly, about lifespans, I think the comparison isn't quite correct?
lifespan seems to be more strongly correlated by size, not squashed-nosed-ness.
Consider chihuahua, shitzu's (and crosses: bichon-shitzu, ...), poodle crosses, heck lagotto (lagotti?). All can live well past 15.
Versus GSPs, great danes, Irish wolfhounds, and so on, coming in closer to say 6-10 years.
I've never really heard argument on lifespan of pugs et al versus other dogs, though. More around (perceived) ugliness/prettiness, and their breathing issues.
There is this wonderful concept of "flow state", which I'm sure most of us are familiar with. It was originated by Mihaly Csikszentmihalyi. Many of us in this thread (and Linus's essay quoted in OP) are referring in various ways of getting into the flow state while programming.
Something I learned late in the game as a programmer is how beautiful it is when multiple people somehow sync up and get into a flow state together. It is like a choir, where everyone is somehow elevated and together they achieve a lovely unity. Or a sports team where all of the players are locked in, seemingly reading each others' minds, and getting to some amazing transcendent place together. The book "The Boys in the Boat" describes this phenomenon beautifully. When all nine members of the team achieve this sort of locked-in unity, the boat just flies across the water.
In software, there are times when ideas seem to pass with electricity from person to person. People immediately build upon the thoughts of others. Ideas seemingly come from nowhere, a synthesis of the conceptual pieces held by each of the team members. You realize that some wonderful new way of doing things has emerged from the dialog, and no one person could have possibly gotten there on their own.
There is a wonderful old-fashioned phrase that describes this:
"The whole is greater than the sum of the parts."
My programming brain has two modes. One is playful mode, and the other is drudge workaday mode. 9-to-5 Monday through Friday is workaday mode. Other times are playful mode. As it turns out, I am a far better programmer in one of those two modes. It won't be hard to guess which one. In drudge mode, I may spend three hours some afternoon trying to force my way through a problem. No turning back, the clock is ticking! In play mode if I spend three hours goofing around with the exact same problem and it's kinda not happening, well, screw it. That was sorta fun, but let's goof around with it some more and try something else. BANG! Sweet clean obvious beautiful solution straight from Erdos' BOOK. The powerful sense of joy in the latter mode sustains my love for programming, but I wish I could convince my lizard brain that it would be OK, indeed far better, to stay in play mode and not get sucked into the dark doldrums of drudge mode during "job" time.
I’m convinced that going for a walk, or staring out of a window, or browsing Reddit, or even taking a nap to let your brain work is an essential part of programming work and that should be understood by any employer. And employee.
May I ask approximately how old you are and what field you're working in? (Or, being passionate in?)
I feel like this is not a realistic view to sustain in most modern tech environments, unless you love inefficiently producing ineffective solutions that just so happens to be profitable, or you job hop every 1 - 2 years.
Interesting, thanks! I feel like that kind of confirms my position: you're not someone who has the same passion for coding, you have a stronger passion for building, leading and selling tech companies, which is why you've done that multiple times.
I've never been a founding engineer, but typically a direct report to founding engineers.
The people I worked with in your position were either like you (e.g. enjoyed programming but clearly got a big rush from the business and money side too) or they were genuinely just pure passionate programmers and miserable, as their role takes them away from that.
There's a whole categories of skills, abilities and problems that are never confronted if a person changes job every <2 years.
Hot take: what if the rise in enshittification and crap tech is because good tech can only be produced with the hindsight of a stable tech career from a stable tech employer?
That's a different topic though. This was about passion in coding for living.
I don't think it's the employee's responsibility to stay in one company if the stability and in-house career path options are questionable, as usually is the case.
If your job is your passion, and your downstream customers give you freedom to treat it as such, it can be fun.
The fun gets sucked right out of it when you have people breathing down your neck waiting for your output, are very particular about what they expect from it in a way that doesn't align with your creative values, etc.
Not all of us have a Lord Saatchi willing to bankroll whatever our brains fart out and call the result brilliant (likely to pump up its value to buyers). Matter of fact, that may just be what ZIRP-era VC can be conceptualized as: business model "Uber for Lord Saatchi-style patronage in tech".
I heard an interesting talk recently by a Stanford neuroscientist.
He was studying standing waves of brain activity among circularly linked groups of neurons. Neurons can provide both excitatory inputs and inhibitory inputs to other neurons.
For a computer programmer, it is easy to imagine excitatory and inhibitory localized phenomena giving rise to all sorts of interesting and complex self-sustaining standing waves.
Think of the study of cellular automata by Stephen Wolfram and others, in which various simple localized rules give rise to all sorts of interesting computational phenomena, up to and including Turing Completeness.
In animal models such as pigs, these standing waves can be observed to persist for months if not longer.
His particular area of interest is to subject the brain to gentle sub-lethal doses of radiation treatment in order to change selected standing waves.
My thought is that there may be a qualitative difference between the underlying "neural hardware" and the thought processes that "execute" on this hardware.
It is the bread and butter of computer scientists to reason in a world in which the complexity of software greatly exceeds the simple computational artifacts on which it runs.
One might say that chess has information content that is qualitatively dissimilar to the wood-working needed to make a pretty chess board.
The information content in DNA is composed of chemically interchangeable A, G, T, and C molecules. So, one might say that the underlying physics is "walled off" from the information content of the DNA. Evolution has, as it were, a free hand to encode advantageous alleles, with no bias introduced by the underlying physics or chemistry.
All this is to say "Hear hear!" to this excellent article.
The "metaphorical brain talk" the author describes may indeed be a conceptual limitation if applied too broadly to the mind.
I'm not really up on neuroscientific research but I'd think a way of progressing might be to make computer simulations of neurons and then see if you can get those to exhibit similar phenomena as those waves and other experimental observations.
Then, I can reason through the AI agent's responses and decide what if anything I need to do about them.
I just did this for one project so far, but got surprisingly useful results.
It turns out that the possible bugs identified by the AI tool were not bugs based on the larger context of the code as it exists right now. For example, it found a function that returns a pointer, and it may return NULL. Call sites were not checking for a NULL return value. The code in its current state could never in fact return a NULL value. However, future-proofing this code, it would be good practice to check for this case in the call sites.