I wonder if there's going to be a mythical AI month.
Adding AI to something makes it later/worse/slop
Mythical Man Month
201–210 of 219 posts
Re: Mythical Man Month
#202Earlier quoted context omitted.
10x what? 10x revenue? 10x features shipped? Whats the measure, is it 10x speed of dev like parent comment? Because an unqualified 10x could mean 10x SLOC which is trivial with an agent but has negative value. Assuming 10x on the speed of dev, Is the vscode repo a decent example? Recently they've been all in on AI augmented development so i'm thinking they'd be a reasonable subject? How do you isolate out what counts…
Its 10x code generation with .5x quality at best and all other parts of the SDLC are at 1.x or worse. AI is not delivering 10x shareholder value, anywhere. Software developers have quite the level of hubris about how important they are to companies. Yes our work is very complex and takes a certain mindset to do it well. It takes a lot of other roles to have a successful business, many of those roles will use AI to he…
> AI is not delivering 10x shareholder value, anywhere.
A bit facetious, but I'd expect Nvidia and the like providing the "AI equipment" to have a 10× share value at least…
Re: Mythical Man Month
#203I re-read that book every 10 years and try to think carefully about whether what Brooks wrote still holds. The last three times I read the book, everything held. This time, I'm not so sure: AI does change things significantly. Perhaps not for all teams and not all scales of software, but in my case (solo developer, complex software system) I did measure a 12x productivity increase [1]. Also, some of the problems Broo…
Dario Amodei said in the most recent interview with Dwarkesh that Anthropic currently gets achieves an increase of around 20-30% coding productivity, which tracks with my experience. What do you do to reap orders of magnitude more? Also, how much more money do you make? Or are you working less?
I can clearly see, and feel, Dario's associates' "increased productivity" in their Claude Code/Chat/Cowork desktop product...
So many updates, pretty much daily, so many tweaks to the interface. Sometimes the tweaks are a bit dumb, sometimes completely trivial, and other times they just undo what they did previously.
My favorite examples of their newfound velocity:
1. When there is a nice feature that was easily discovered, and then it was gone...
2. When the "customize" section moves around to random places in settings, or entirely out of it
Special mention to their scatterbrained keyboard shortcuts strategy
Re: Mythical Man Month
#204Earlier quoted context omitted.
Life has never been more affordable than it is now. Virtually all of your ancestors were impoverished to a degree you can't even imagine.
Life was becoming increasingly more affordable, but that stopped being the case years ago. It is now declining. I would like it to either stay the same as it was years ago or start increasing again.
Sure but it’s still more affordable than it was for the vast majority of our ancestors, and they still had kids, so it’s not an explanation for declining TFR in developed countries.
Re: Mythical Man Month
#205Notably, his essay “no silver bullet” states that there has never been a new technology or way of thinking or working that has led to a 10X increase in the speed of software development. That was true for almost seventy years until roughly last year. AI is the silver bullet - my output is genuinely 10X what it was before claude code existed.
I haven't yet seen anyone with a concrete example project (public ideally, but even describing private efforts in enough detail to enable potential criticism would be fine) making a claim as strong as 10x. Are you willing to break the mould and show us what we're all missing?
I have a programming language project: lone lisp. It's on GitHub. I asked Claude to analyze the commit history before and after AI and the result was: with AI I'm committing around 8x faster than before. Lone experienced a literal explosion of features, improvements and fixes after I subscribed to Claude.
And this is the project I care most about. I review every single line of code the AI outputs. I push back on everything it does. I reword every commit message. I make every effort to understand how everything works before committing it to master. I cook branches for days and I don't merge until I think it's perfect.
Still gave me an ~8x improvement.
The latest development: shaped objects, like Self and V8. I asked Claude about it and it just implemented it in like 10 minutes. Boom, instant ~20% speed improvement. I read the code and it turned out to be almost obnoxiously simple. It basically converts hash tables into arrays internally, and deoptimizes back to a hash table if anyone deletes keys from it. I'm still reeling from the sheer absurdity of it.
Re: Mythical Man Month
#206Earlier quoted context omitted.
Just create a skill for it -> I call mine `babysit`. It spins up a subagent that polls it every x minutes and auto-fixes it until it's green. I already continue with the next task while it does that in the background
I do this with our AI PR review checks. We have AI review every PR and commits to PRs... which can cause long running loops of commit fix. So my agent just listens for green checks and no PR comments and loops until those conditions are met.
About 30% of our AI PR review checks are flawed at a fundamental level, based on poor comprehension of the whole. If I told the AI "fix everything and keep going until it's green" I would be terrified of the result.
I disbelieve this works in anything other than a toy codebase (or an incredibly fine-grained microservice).
The 70% is amazing! But a 30% failure rate requires intense supervision.
Re: Mythical Man Month
#207so much bad software was written because of this guy (Martin) selling snakeoil
Re: Mythical Man Month
#208Earlier quoted context omitted.
My friend who has never taken a programming class (or even touched an IDE before AFAIK) has now put a small app into production. Complete frontend + backend + database. Yes, it is an internal app, but it works and everyone loves it. Does that count as an example? (Also I absolutely expect him to need help at some point, but so far it has taken his project from absolutely impossible to 3 weeks of work in between work,…
My friend who has never taken a programming class did the same 10 years ago.
Obviously it was not the same s since models that could write programs were not available 10 years ago.
And if you just mean your friend taught himself programming on his own, well that is actually very cool, I did too back in the 90ies and so did many others here.
My point is that it is now possible to vibe code a full application from frontend to backend today and still not be able to understand a line of TypeScript or anything else.
Re: Mythical Man Month
#209Earlier quoted context omitted.
My friend who has never taken a programming class (or even touched an IDE before AFAIK) has now put a small app into production. Complete frontend + backend + database. Yes, it is an internal app, but it works and everyone loves it. Does that count as an example? (Also I absolutely expect him to need help at some point, but so far it has taken his project from absolutely impossible to 3 weeks of work in between work,…
These kind of things (internal tools created out of band of normal engineering practices by non-engineers) were amazing back when I did pentesting because the security was always the last consideration. That got harder when SaaS became preferable to rolling your own stuff for everything. Guess things are gonna get fun again for red teams lol
I agree with you.
The danger is not however that only that people write their own tools for calculations and capacity planning etc.
The danger is people make useful stuff that is very fine as long it is just an internal tool, but then someone add credentials to other systems so it can access and maybe even update stuff and it gets exposed to third parties etc and all of a sudden we have a major data breach going on.
Re: Mythical Man Month
#210Notably, his essay “no silver bullet” states that there has never been a new technology or way of thinking or working that has led to a 10X increase in the speed of software development. That was true for almost seventy years until roughly last year. AI is the silver bullet - my output is genuinely 10X what it was before claude code existed.
I haven't yet seen anyone with a concrete example project (public ideally, but even describing private efforts in enough detail to enable potential criticism would be fine) making a claim as strong as 10x. Are you willing to break the mould and show us what we're all missing?
It's often much more than 10x, because AI makes things so much faster and easier that we are now doing many things we wouldn't have bothered to do at all before. And there's no reason to believe we're near the finish line yet as to how good LLMs can get.