Hacker Newsnew | past | comments | ask | show | jobs | submit | risyachka's commentslogin

Pretty sure at this point they train it on pelicans

In any usable product making a product is like 20% or less. Enter compliance, security, payments and a million other things.

Even if you can build it in a day B2B SaaS will continue to prosper because they sell peace of mind, reliability and compliance. Not features.

Also due to economy of scale it will always be cheaper to buy something from a vendor that sells it to many clients than to DIY it.


Yep. Wait until they realize one of their vibe coded apps validates everything client side and their entire DB is open to the world.

Like that never happened with non-vibecoded stuff.

Of course it has, but it's more likely to happen when nobody is paying attention.

There's a difference between someone breaking into your house to steal your valuables, and you getting robbed after leaving your valuables on your lawn with a sign saying "Look At My Valuables"

Startups of the last decade were routinely doing this. Building their fresh ideas on the newfangled databse without bothering to investigste how to secure it in any way against malicious actors.

Yep. It's a funny thing.

You build a Twitter. Profiles have posts, posts can have images, etc. It's very easy to model the database.

But then how do you make money with it? Now you need to build a separate system for advertising? Or do you want to sell subscriptions? Which means you need to build a separate system to handle payments. This is usually the big one, because when you handle money, what happens if there is a bug and you charge someone without delivering anything? How do you prevent fraud? How do you handle disputes?

Someone posted something illegal. What do you do in this situation? Do you call the police? The FBI? What kind of data do you give the authorities? How much data SHOULD you have been logging in the first place in case something like this happens?

One user doesn't like you so he bought a botnet to DDoS your website. How do you handle this? Are they mass posting? Mass creating accounts? Is it possible for them to exhaust all the usernames possible and then nobody can create an account anymore?

Your website is online but if the server blows up you'll lose all the data in the database. You need backups. You need a system to ensure the backups are actually working. But then some guy from the UK said he wants his posts all deleted. What are you going to do now, because his posts are also in the backups, and you don't want to touch those.

Trolls are posting things against the ToS. Who handles these things? Shadowban? So there needs to be a shadowban system? Moderators? So there needs to be a moderator-only section of the website? Should this be integrated with the main website or not?

Then you look at this horrendous mess of 6 paragraphs and you think back about the first paragraph that already did everything you wanted from Twitter. All these other systems, most of the work, and all you actually wanted was the first paragraph.


All those things are true. It still doesn’t sound like 1000+ engineers at 350k/yr.

What actually happens in a startup is you encounter these problems one at a time as they arise.


Twitter wasn't built by 1000 engineers at 350k/yr.

It had to hire them later on. Because when there are users - you need support, take out fires etc.

And this exact thing will happen with any homebrewed SaaS.

You either run a business or play tech company making your own saas instead of focusing on your business.

Sure you can do both in very rare cases - if you are SpaceX or similar, otherwise you are shooting yourself in the foot.


No. They hired 1000 people to help them justify funding rounds.

Even then, startups prioritize growth over efficiency. So maybe 100 people would have been fine but 1000 gets them a 5% growth improvement in growth.


Startups have no users and no data to start with, and if they fuck up security, well, they just fail sooner than expected.

Once you get past a certain size, you have very different sorts of problems. Any idiot can vibe code a facebook lookalike, but the real one has to handle hundreds of millions of users and posts while being a target for state actors.

TLDR; yes you do need that many


> yes you do need that many

You absolutely do not. what do you think about the website we are using right now! It has half of the problems listed above.

> facebook

Your work project doesn’t have a billion users.

We were talking about what it would take to fix the technical problems resulting from taking a working program to something people use.

> Any idiot can vibe code

I didn’t say that either.

How is it the HN opinion that it’s impossible to make a web application a lot of people use?


These are two different problems.

AI has solved the 'coding' part. The business is still very human because they are the ones buying, for now.


It's still possible. Have the product and hire other people for these things.

Use stripe, cloudflare, whatever the legal equivalent of these stuff, S3.

Yes they might take most potential profit, but you'll also not have a huge payroll.


You do realize that making software by developer for his wife means for random business is hiring a third party dev to build custom software?

So still, for ransom business much cheaper and better to buy software from SaaS vendor.

In this case it was better ONLY because the client is the wife of the developer.

And even now, if he sells this to other businesses - it will be MUCH cheaper to buy his subscription than homebrew the same version of it - as if it starts selling it he will be adding features and support which requires time (which is money).


> You do realize that making software by developer for his wife means for random business is hiring a third party dev to build custom software?

No, this is not true. There are so many non-technical users of Microsoft Access that run their won businesses without hiring anyone. A friend of mine had a business with an yearly turnover shy of $3M (which is small, alright) and it was running wired spreadsheets and google forms. 20 people. He never ever bought any software, and existed for more than 10 years, until his wife (yes his waifu) decided to divorce and bring the company down.

Business Architecture is not so much about writing the software, sorry, we as IT professionals would love to think it is, but this is a super weak bias.


Almost every SMB I interact with sounds like this company. Was it founded more than 10 years ago? Probably holding the ship together with spreadsheets and email.

Yep, and there are so many of these examples, as you said - most SMBs. But I know also more recent companies having it similarly.

On the other hand - we did recently pitche some brand new RAG for TBs of complex schematics for a company in the Netherlands, and guess what - they didn't like the enterprise rates that the middleman offered, not because they did not like the demo (which they loved absolutely), and not because it was late or incomplete (it took less than a month). Had I approached this company directly, with normal rates, I would be deploying it in production already. It's very telling, and not good news for large SAAS vendors.


So true, we devs are so stuck in our own bubble

>> He never ever bought any software

>> was running wired spreadsheets and google forms

So he was using someone's software.

Probably he grew his business to 3M for this exact reason - by focusing on business and not playing tech company by homebrewing software. Which only proves my point.


Outsourcing thinking is not a skill. It is the same as skipping gym. Nothing to practice here

A lot of people practice not going to a gym! I bet it reflects e.g. on their dating outcomes, at least statistically.

I suspect that outsourcing thinking may reflect on quite some outcomes, too. We just need time to gather the statistics.


All websites had all incentives to be api-first for 2 decades now.

They aren't API first not because they can't - but because customers will use whatever GUI website supports to use it.

Unless your product has no value - users will fill out those forms manually if you don't provide API if they need your product.

Agents don't change this.


They do, for the simple reason that I don't want to have to click through forms and pages and hunt things down manually to do a workflow. I won't use a cloud provider with a broken or half-functional cli for this reason, and if anybody in any space provides a better agent interface for doing stuff so I don't have to click through UIs, I will switch pretty much instantly.

Think of agent support sort of like food delivery. Once people get used to the convenience, places that don't offer it are at a huge disadvantage.


>> I won't use a cloud provider with a broken or half-functional cli for this reason

You will if they have what you need and others don't


>> AI will eat the "software professionals"

you mean AI will eat everyone, because if software professionals will be automated - all other white collar jobs will be too via software.

And then all resources will be poured in hardware and blue collar jobs will be automated too, at least those that have more value.


That’s the thing here. Software engineering is an intelligence-complete problem. If AI can solve it, then it can solve any sort of knowledge work like accounting, financial analysis, etc

Only if by "solving it", you mean being able to write any program to do anything.

Software engineering is a hubris-complete problem. Somehow, being able to do so much seems to make us all assume that everyone else is capable of so little. But just because we can write 1000 programs to do 1000 different things, and because AI can write 1000 programs to do 1000 different things, it doesn't mean that we can write the million other programs that do a million other things. That would be like assuming that because someone is a writer and has written 1 book, that they are fully capable of writing both War & Peace and an exhaustive manual on tractor repair.

Financial analysis is not easier than programming. You don't feed in numbers, turn a crank, and get out correct answers. Some people do only that, and yeah, AI can probably replace them.

"Computing" as a field only made sense when computers were new. We're going to have to go back to actually accomplishing things, not depending on the fact that computers are involved and making them do anything is hard so anyone who can make them do things is automatically valuable. (Which sucks for me, because I'm pretty good at making computers do things but not so good at much of anything else with economic value.) "What do you do?" "I use computers to do X." "Why didn't you just say you do X, then?" is already kind of a thing; now it's going to move on to "I use AI to do X."

Then again: the AI-dependent generation is losing the ability to think, as a result of leaning on AI to do it for them. So while my generation stuck the previous generation with maintaining COBOL programs, the next generation will stick mine with thinking. I can deal with that. I like thinking.

</end-of-weird-rant>


> Financial analysis is not easier than programming. You don't feed in numbers, turn a crank, and get out correct answers

It’s not, but if software engineering is solved then of course so is financial analysis, because a program could be written to do it. If the program is not good enough, then software engineering is not solved.

I think this what you were getting at with this part, but it’s not clear to me, because it seems like you were disagreeing with my thesis: “ because AI can write 1000 programs to do 1000 different things, it doesn't mean that we can write the million other programs that do a million other things”

I’m not sure if you’re saying that people weren’t using computers to solve problems before, but that’s pretty much everything they do. Some people were specifically trained to make computers solve problems, but if computers can solve X problem without a programmer, then both the computer programmer and the X problem solver are replaced.


I don't think software engineering is ever going to be solved, but financial analysis will definitely never be solved. It's impossible, the nature of it dictates that, whatever changes happen will further change the results. Financial analysis requires novel thinking, and even if you have AGI that can engage in novel thought they will just be another input into the system.

Just like AI, the winners will (continue to) be the ones with the most access to data and the technical and financial capital to make use of it.

This is the crux of it. The digital world doesn't produce value except when it eases the production of real goods. Software Development as a field is strange: it can only produce value when it is used to make production of real goods more efficient. We can use AI to cut out bureaucratic work, which then means that all that is left is real work: craftsmanship, relationship building, design, leadership.

There are plenty of "human in the loop" jobs still left. I certainly don't want furniture designed by AI, because there is no possible way for an AI to understand my particular fleshly requirements (AI simply doesn't have the wetware required to understand human tactile needs). But the bureaucratic jobs will mostly be automated away, and good riddance. They were killing the human spirit.


> Software Development as a field is strange: it can only produce value when it is used to make production of real goods more efficient. We can use AI to cut out bureaucratic work, which then means that all that is left is real work: craftsmanship, relationship building, design, leadership.

Thats a really odd take. Software is merely a way of ingesting data and producing information. And information often has intrinsic value. This can scale from simple things like minor annoyances of forgetting your umbrella, to avoiding deaths/millions of dollars in losses due to ships sinking in storms.

Now the long term value of software does approach zero, because it can usually be duplicated quite easily.


Extraction and manufacturing are considered the primary and secondary economic sectors. In a closed loop system, tertiary and onward sectors, like services and technology, cannot exist without the primary and secondary.

I value your weird rant. Yes it did go on as a thought stream, but there's sense in there.

I've been thinking a lot around a kind of smart-people paradox: very intellectual arguments all basically plotting a line toward some inevitable conclusion like super intelligence or consciousness. Everything is a raw compute problem.

While at the same time all scientific progress gives us more and more evidence that reality is non-computable, non linear.


> While at the same time all scientific progress gives us more and more evidence that reality is non-computable, non linear.

What scientific problems are non-computable?

ANNs are designed to handle non-linearities BTW, thats the entire point of activation functions and multi layer networks


non computable, non-linear as in given known input parameters you can determine the output parameters.

we can't do that for mostly any complex physical system, as would be for something like living organisms.


> non computable, non-linear as in given known input parameters you can determine the output parameters.

These two words do not mean the same thing.

Non-linear functions do not mean you cannot determine the output for a given input.

All non-linear means is that the condition f(x+y) = f(x) + f(y) and f(kx) = kf(x) do not hold for arbitrary x,y,k

For example f(x) = x^2 is a non-linear function. Can you determine what f(x) for arbitrary x?

Perhaps you meant what used to be called "chaotic systems", those which were highly sensitive to initial conditions. Yes, they are non-linear but they are completely deterministic. A classic example would be the n-body problem in physics under most conditions.

And I'm not sure what you understand what non-computable means. It means that the computation will not halt in a finite amount of time for a general input. For a particular input, it may indeed halt in a finite amount of time.

Most real numbers are non-computable, such as the square root of 2 or Pi.

Practically speaking however, we can get approximations as close as we want. In other cases, such as the Busy Beaver function, we can set bounds


You're correct. I only have a very casual understanding of these things. For the non-linear thing, I just mean that for any advanced system there are say trillions of parameters, like cellular systems, and even if you mapped them in you couldn't be sure what the output would be.

    > And I'm not sure what you understand what non-computable means. It means that the computation will not halt in a finite amount of time for a general input. For a particular input, it may indeed halt in a finite amount of time.
Sounds familiar, the "halting problem"? I suppose I'm too loosely tying concepts together. Particular vs general input is same as simple vs complex input above, given a complex enough input, the compute involved approaches boundless/infinite.

In practice, yes, as I understand it, modern science is all about stochastic approximations and for all intents and purposes it's quite reliable.

I probably should stop using "non-linear" terminology. I really just mean that it's not 1:1. You mention how systems can be deterministic and I looked it up and yes wave function collapse specifically says:

    > The observable acts as a linear function on the states of the system
We can compute the possible states, but not the exact state. We can't predict the future.

Thanks for the reply, this is much more interesting to me as it approaches philosophy, so admittedly I too loosely throw words-that-mean-things around.


You are right, but I think at the moment, a lot of people are confusing "software engineering" with "set up my react boilerplate with tailwind and unit tests", and AI just is way better for that sort of rote thing.

I've never felt comfortable with the devs who just want some Jira ticket with exactly what to do. That's basically what AI/LLMs can do pretty well.


Those people have always annoyed the hell out of me and I would prefer to not work with them.

You're right. I think the current AI direction is a dead end for real artificial intelligence, so it is not the thing that will replace all jobs, but the day a machine with the real cognitive capacity of a 5 year old exists is the day almost all of humanity becomes useless.

And before that the current direction is still enough to massively hurt the world because there will be less and less places for us humans.

Another point I noticed that nobody is talking around us is the technology adoption rates. When the car industry started, decades happened between the early users and cars being ubiquitous in the population (especially taking into account the world and not the richest countries). So a sizeable part of the transportation industry that was ultimately replaced by cars had the time to adapt, move to other jobs or arrive at the end of their work life.

But now the technology goes from its few first users to being used by everyone and their cats in years if not months. There is absolutely no time to adapt, love over or endure things until you don't work anymore.


Software was already at its limits on automation, the last thing automated will be writing code that does the required thing but automating other stuff that wasn’t already automated by software will take some time because will require AI advances in those particular domains.

Once an AI runs a single company well, all publicly traded companies will have a legal obligation to at least consider replacing the C-suite with AI. In theory. I'll believe it when I see it.

Thats not cheaper than paying a subscription. In fact this is at least 3x-10x more expensive.

And this is comparing to being subscribed many years in a row. With SaaS you can unsub and sub only when you need it again.

With your side project - a weekend of your life is invested and you will never get it back.

This is the worst use of your time if you measure it in $. If you make it for fun - sure. In all other terms it is a complete loss.


$20 claude code subscription for a month can replace the $15 + $10 for each month. How is that 3x more espensive? The user just saved $280 per year, on just two subscriptions alone.

Hardly doubt that this was the 'most waste of ones time'. For one, it's not like most of us can decide to "work" for 3-5 hours on a Saturday and get any money. I play games on my pc while claude codes for me. I alt tab each few minutes and see if it needs any input. Then I can (not that I do it), read and perhaps learn from the code.


Yup. thats more expensive because each hour of your time is at least 50USD And each hour on weekend that you would have spent with your family etc is probably 500-1000 usd at least, so yeah, it is much cheaper to pay 15 usd for SaaS


Hey, CC has a way to trigger a notification chime on completion.

(How? Idk, I just asked it to guide me through the short hook process)


Afaik it's built in now.

I get a notification on macOS with the title of the context.


Yup. This.


>> But despite all that, I’m still having a blast

This is proves its mostly over for the high income industry.

There are no good paying jobs where you are having a blast. Otherwise there is a lot of those who want to do that job which drives wages waay down.

High paying jobs are tough/stressful/not fun. Which was the case with software before.


What if I'm having a blast doing the tasks no one wants to do that are in the backlog.


Maybe you're incredibly lucky? But I wouldn't count on that luck lasting.


Bullshit. There are some dysfunctional exceptions, but higher pay usually correlates with more freedom, more responsibility and access to more resources. You may get frustrated in Big Tech, because nowadays, people like to have no bounds in their expectations, but try changing your big tech job for a job in the oil or insurance industry with half the pay, and you'll see that most of the time, you not getting a blast out of your 600k job is mostly your fault.


>> Just so long as you have tests that show ...

this by definition filters out all non-devs, even many junior devs as you need to understand deeply if those tests are correct and cover all important edge cases etc.

+ when you deploy it - you need to know it was properly deployed and your db creds are not on frontend.

But mostly no one cares as there is no consequences to leaking personal data of your users or whatnot.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: