dustingetz

the leader isn’t dancing because they want to dance, they are dancing because the people want to dance

forget all that write a google doc make it accessible and clear post on relevant social media probably twitter let the people decide

determinism isn’t really enough, we want “predictable”. Most of these AI wavefunctions are “chaotic” - tiny changes in state can cause wildly divergent outcomes

the essence of man is blind spots and hubris

i would add that the reason no product manager wants to pay for #3 is because historical attempts to do so have overwhelmingly resulted in cost/schedule overruns; did-not-finish outcomes are common. Let he who believes otherwise demonstrate so with his own money, this is called a startup and note that virtually all startups fail i.e. run out of some critical resource without finishing! So what is a wisened product manager to do? No easy answers here - simply look to the industry to see what the average outcome is. And it is not for lack of trying. in my opinion software delivery is not a solved problem. but it is really hard to make money as a software delivery expert by going around and saying that you don’t know how to deliver software.

data engineering was free money for nothing at all circa 2014, they got paid about 1.5x a fullstack application developer for .5x the work because frontend/ui work was considered soft, unworthy

do you think it’s really exponential in practice? i don’t know if i see that, and polynomial (growth in bugs per LOC) is enough to explain why all sufficiently large projects grind to a halt

use git subtree - first to concatenate the minor repos into one major repo, and then subtree split from that point forward to publish subtree folders back to the minor repos, if needed (e.g. open source projects to github). works for us with about 8 minor repos, eliminated submodule pain entirely. only the delivery lead has to even know the minor repos exist.

it exists under superdeterminism, which is the most sane interpretation of wavefunction collapse

humans cannot in any sufficiently large project, for example government contracts and the entire banking sector.

The investors may be right, and yet, this is a collapse vector in “The Collapse of Complex Societies”, in one sentence, “societies collapse when they reach a point of rapidly declining marginal returns on their investments in problem solving capacity.” So we’d expect an outcome somewhere between “market dynamics will resolve this” and “war and/or violent revolution”

1/ this requires AGI not stochastic parrot - as the training dataset does not contain any successful instances of obtaining actual DRY at scale, so the agent will need to contribute novel research and essentially solve software engineering. It’s not 2-3 years out. And why should any result be aligned to how things work today, why not start over from the qbits and work our way up?

2/ The worst instances of coupling are between foreign systems where one is not in your control to change, or too high risk (such as banking) and prohibitively difficult to test all but the most incremental changes. So at some point you need to decide what error rate is acceptable in any tool assisted rewrite, where error rate is measured in death count and lawsuit liability.

If the cost per unit delivery (say LOC) grows something like polynomial due to interconnectedness, surely it is superlinear for even a state of art formal reasoning system let alone imperative langs, it will outstrip our ability to build energy infrastructure. If the cost growth is exponential due to state space explosion, it will outstrip the energy in the Sun.

How does HN think these factors will balance?

yeah, i think the siren song of reducing developer cost and accelerating delivery is irresistible to boards, execs, managers, non-senior devs (all essentially semi-technical, even the nonexpert devs) but it is a mirage. The startups that over raised in the extended bubble are all drowning in tech debt and the game is to exit or get profitable faster than the tech debt compounds, it leads to a divergence in outcomes where you either build a google and buy your way out of the hole (“ revenue solves all problems”) or you don’t quite get there and your delivery ability crests and decays as the tech debt compounds above the threshold line that the company cash flow can service.

Vibe coding is going to exacerbate this - the growth can come faster, but the collapse point is accelerated correspondingly. If you plot this complexity out like we were taught in CS 101, lines of code will grow linearly, proportionate to the number of agents/devs, but the interconnectedness and therefore state space will explode polynomially without applying encapsulation and local reasoning techniques like functional programming. But we’ve aligned the models to use imperative languages (TS, Python, Java), essentially constraining the code models to specifically prevent the best code scaling techniques! And guaranteeing the emergent “spooky action at a distance” that is inherent to the computational structure of imperative languages.

And note that most companies don’t take flight like Google, instead they seek acquisition, and a lot of those acquisitions fail to integrate and get swimming, they are perhaps underwhelming compared to the sales pitch, to the extent that many start to look like a form of securities fraud? Tech debt is real, it’s just invisible to the balance sheet and therefore to boards and buyers. If it were properly accounted for, the entire IT sector would be revealed as completely underwater.

Time Is The Denominator,

especially in an industry context where you are being paid (there’s that pesky denominator again). R&D, which is what this is, is a very expensive and risky endeavor. 99% of startup projects fail, for example. They would not fail given infinite time, but you cant have that because time is not free:

Interest Is The Price Of Time

$30M Series A by Tiger in 2018, acq by Fastly in 2022 for undisclosed price, shutdown in 2025. Fastly is a CDN and edge network. Glitch was used by Fastly for “internal training” and “internal sales tools” (kinda?), a far cry from “fullstack application platform in the edge”.

https://blog.glitch.com/post/my-last-day-at-fastly/

https://www.fastly.com/blog/fastly-announces-acquisition-of-...

“I contacted all the Vcs in the world, twice” cold outreach can work but the email needs to be an absolute bullseye to the exact VC in the world who already has your hypothesis and wants to learn more. But even if it’s love at first sight, you still need momentum with other VCs to get them to pull the trigger, which means you need interest from generalists, which means you need a few dozen warm intros. Broadly intros at seed stage come from angels who have already invested or are close to investing. Getting first checks from angels is a less hard game (because they likely have deep technical knowledge of your space and can therefore recognize the brilliance of your team and product, see the staggering opportunity you are after w/o you needing to baby-splain it) but it takes a long time, years, to reach them, as no “ordinary human being” wants to be hit up by cold emails from a cold startup asking for money, in my experience they need to come to you inbound via your brand and network. Using twitter DMs to meet as many people as possible (once your reputation is established) is a good way to get networked and start to map out who’s who in your space and develop the alliances you will need to succeed, get intros to customers, eventually get intros to acquirers (even harder game than VC, acquirers are even less technical and more spreadsheet-brained than VCs)

you’re playing a very competitive game, i recommend working with a coach, many on linkedin with accelerator experience it will cost you about $5-10k and you’ll get accurate feedback about where you stand. i spend more than that on advisors each year. raised a mil need more

if this is zero sum-ish then who is the counterparty that’s getting screwed and how?

well, yes, because the imaginary world has no real world constraints

NYC and SF have private capital concentrated in enormous quantity. Look to the preceding 5 decades to see how that happened, it did not spring up overnight.

do they also watermark the code?

Electric Clojure https://github.com/hyperfiddle/electric

"Time is the denominator"

who said anything about LLMs? Did Pichai specify that?

it also drives my gmail sentence autocomplete. it does not mean “30% of my email is written by AI”. It does help me type faster though. Reframing the one as the other is, imo, securities fraud. (I will asterisk that YC startups vibecoding their product is real, but that’s, like, 10^10 lower LOC scale than “all of Google”)

this was reported as actually just IDE tab completion back when google claimed this stat last year

i dont understand, is it doing your schoolwork?

it’s a tech culture war, devs are gaslighting the business sucking down incredible salaries while reporting that the jira ticket is delayed by tech debt, the business is gaslighting the devs with the inept technology leadership, stupid HR games (my wife’s workplace dragged everybody in for fucking friendship bracelet day), secret DEI policies while paying high performers the same as negative performers because they can’t tell the difference.

These two groups hate each other and AI promises holy grail to both parties - devs can more easily learn the 1000th tech stack in between yoga classes while pretending to work, the business dreams of finally firing all the devs and keeping all that money for themselves. And neither group has a clue how this dream will be fulfilled, but they want to believe because computers can now talk, so enter the entrepreneurs and VCs to gaslight everyone involved with fake stories of how 35% of LOC at Google was coded by AI (erm, accepted IDE autocomplete), laughing all the way to the bank while they vacuum up all that dumb money, poor befuddled executives that roleplayed their way into an 8 figure budget responsibility by being tall white men with blue eyes

TSA theatre, america fuck yeah