byoung2

As with most technological revolutions, some jobs will be lost, and many more will be created. The invention of the automobile killed the horse and buggy profession, but it created car manufacturing, road construction, petroleum, motels, long haul trucking, etc. There is potential for AI to enable people to do jobs they could not do before. There are millions of people who could say this sentence: I would love to do X job but it requires Y skill that I don't have, so I do Z job instead. AI is quickly allowing everyone to do Y skill, so it is unlocking a lot of X jobs for people currently stuck doing Z jobs (or no job).

I'm old enough to remember similar talk of the internet destroying jobs. Phone information operators will be out of business because websites can give you better info, faster. Print media would disappear. Travel agents and realtors would be replaced by websites. Some of that happened, but many jobs were created. Web developers, cybersecurity, content creators, and hundreds more jobs that no one dreamed of now exist because of the internet.

In some ways that “problem” is an escape hatch for Disney. Other studios have to write off huge losses for underperforming movies (and Disney definitely does this too). But even a flop can drive Disney+ subscriptions, park visits, merchandise purchases, hotel stays, cruise bookings, etc. Star Wars is an excellent example, even though many of the Disney era movies underperformed (it’s hard to say flop), Disney has made back more than the purchase price of Lucasfilm, the budgets of all movies and shows, and construction costs of 2 Galaxy’s Edge lands and turned a profit. They’re like Costco in that way where the food court doesn’t need to profit as long as they sell memberships.

I recently left Disney after working there for almost 5 years. It's much worse on the inside than people realize. Very sad indeed.

I haven't read the new book yet, but I'd be interested in your take on Disney's trajectory over the last, say, 2 decades. It seems to have strayed pretty far from Walt's original vision, largely due to the actions of Bob Iger. He took what used to be a company that was fueled by creativity and turned it into a machine that strip mines IP and extracts value. Iger purchased IP (Pixar, Lucasfilm, Marvel, Fox) as a risk mitigation strategy since you get an established brand you can exploit on day 1. But in doing so he killed the soul of Disney, which was built on big creative bets (literally sell the car to make a movie, mortgage the house to build a park).

Yeah I remember the "X for Y" format being encouraged [1]. But X was always an example of a successful startup that was a leader in one category/audience, and you were saying you would be the X for a different category or audience.

PG's examples here were AirBnB as the "Ebay for space" or Viaweb as "the Microsoft Word of Ecommerce".

I'm not rewriting history when I say that the X has changed from a representative example of a successful company to a lazy broad technology like "AI agent for insurance" or "AI native recruiting". Here is a current YC batch startup: Manicule - AI Native Developer Relations. I have no Idea what that is. Does it talk to devs using my product like in a chatbot? Does it help them write code using AI? If they said "HubSpot for developer content" or "Vercel for developer relations" I would get it right away. But better than the X for Y formula would be just to describe the startup: we provide an AI-native developer relations team that owns documentation and technical content end-to-end so you don't have to hire someone for $300k.

  1. https://news.ycombinator.com/item?id=6866822

My exact wording was "fully automated". Yes actively replaced, but cashiers have not been fully automated. That would mean no more cashiers at any grocery store, convenience store, because all they have are self checkout machines. That has not happened except at Amazon Go, and they closed all those stores. So my claim is accurate.

Correct $Foo for $Bar was common back then but I made the subtle distinction between "TikTok for Math Tutors" ([solution] for [audience served]) where naming a startup in a different vertical is a shortcut for the solution, and what we see now which is "AI agents for Banking" ([technology] for [industry]) where naming the technology doesn't make it clear what the problem or solution is, and the industry doesn't necessarily say who has the problem or needs the solution.

When I worked at Disney, there were some jobs where people's entire jobs were compiling reports and following up with various departments. Like taking lists of security vulnerabilities from scans and getting commitment dates to fix them. They would take the data out of one system and put it in a spreadsheet. Then they would reach out and create Jira tickets for the teams responsible and then schedule meetings if necessary to discuss. These roles are definitely at risk.

Exactly, and the formula in the past was less "[hot new tech] for [some industry]" back in the day, and more "[problem solved] for [target audience]". Maybe it is just the wording that I object to, and not the substance of the startup's solution to the problem?

I guess I don't care about today's "AI agent for the agricultural industry" as much as I cared about yesterday's "Tool to help farmers plan crop rotation".

Stepping back from LLMs, look at other jobs that could have been fully automated using technology, but haven't. Some jobs, like a grocery cashier could be automated with self check, or waiters could be replaced with phone ordering. We still have humans in these roles, decades after they could have been replaced.

I agree about the bias to believe the past was better than it actually was. I think it's called rosy retrospection, like when people are nostalgic for their high school days even though they were bullied and had no friends. Maybe HN isn't getting worse, it's just that the negative parts are more recent and noticeable.

The recent batches all use the following formulas which I don't find particularly interesting:

  AI native ______ for _______
  ______ for AI agents
  AI _______ for _______

I agree there. Back in 2009 I used to be excited by each new YC batch. There were fresh new ideas like Dropbox, AirBnB, Instacart. I feel like there were cool stories around those startups, like Drew losing a thumbdrive and coming up with the idea for DropBox, or the AirBnB marketing hack with the cereal, or Instacart's founder having beer delivered during the YC interview. I can't think of any recent YC startups that I've cared about. Maybe it just became saturated or maybe I just moved on to other things. I will say this, that the ideas I had 15 years ago needed a startup and investment. Now I'm able to build a lot of those ideas using the cloud and AI. It seems inevitable that some solo founder will soon be able to build a unicorn with no investment and no employees.

They merged with LiveNation and they own half of the venues. The other half of the venues have exclusive deals with TicketMaster, who provides them with software to run venue logistics (TicketMaster for business), creating vendor lock-in.

Apple has to take privacy more seriously than their competitors since they've marketed it as a competitive advantage. So they've focused more on in-device AI features than cloud-based models.

I would argue that social networks died a decade or more ago, and were replaced by social media. When I joined Facebook in 2006 a big part of it was finding and connecting with friends from school. It was cool to see their vacation photos or kids' graduation pics. But now it seems that friends have been replaced with followers, and instead of peer to peer connections of a social network, we have one to many broadcast of social media.

Kindly revert is my favorite

A founder came to you with just an idea and you went on to build it? You should have been a cofounder from the beginning.

Is that repo public? I see a 404

Two things can be true at the same time: people hate ads, and that they are effective. But more likely the reality is more nuanced. I think that there are a minority of people that hate ads and they are very vocal about it. A majority of people are likely indifferent, and passively consume the ads and then buy the products.

Finding a needle in a haystack? The easiest way to do that is to use a magnet. The magnet relies on attraction that the eyes cannot see. So maybe the best dating app for long term matches would have the pics blurred out until you connected on some other level

As long as you have a reasonable explanation, it shouldn't matter. I have a lot of 1 year stints, but I can explain them:

I was at TrueCar when the pandemic started. Suddenly everyone was working from home and new car production was halted. With fewer people in the market for new and used cars, TrueCar laid off 30% of the company. I survived but left before the next wave of layoffs because we were in lights-on mode, not building anything new. (1 year at TrueCar)

I went to Cameo next, and saw the company become a pandemic darling and a unicorn during my tenure. I helped build new features such as video messaging and subscription fan clubs, but as industries started to reopen, actors, performing artists, and athletes went back to work, along with the general public returning to the office, both sides of the celebrity shout out market began to tank. A left days before the layoff announcement. (1 year at Cameo)

Non-profit is not the same as not monetized. Somehow you will have to pay for the infrastructure running the app. If it's ads, you likely wouldn't have a large enough audience to get enough impressions. If it's subscriptions, how is that any different from other paid dating apps? I'm actually working on a dating app at the moment and it is currently not monetized:

  iOS: https://apps.apple.com/us/app/nectar-dating-by-jubilee/id6476018063
  Android: https://play.google.com/store/apps/details?id=com.jubilee.nectarmobile

3...an odd choice, I know

I have been a member of HN since 2009. Back then I was fascinated by startup culture after seeing major exits like AdMob and Zappos. I used to think there was something magical or lucky about achieving startup success. But what I learned from 15 years on HN is that the secret to a successful startup is pretty straightforward (if not easy). Solve a problem that a lot of people or businesses have, charge money for your solution, talk to customers, stay lean early on, fail early and often, iterate and improve, and don't run out of money.

The description says this:

How this content was made Altered or synthetic content Sound or visuals were significantly edited or digitally generated.

I was having trouble getting GPT-4o to extract data like address, email, phone, tracking number from random emails in an inbox. Sometimes it would do it perfectly and other times it would fail miserably on a similar email. Then I tried asking it to first markup the email with schema.org metadata. Then I asked it to extract the data from the schema.org markup. That worked nearly every time.

Maybe there is an extra step you can work into your prompt that would help it get to the proper classification

Looks like a software glitch, but no confirmation

There is a reason it is called vibe coding and not vibe architecting or vibe systems design. I have had better results when I come up with the overall architecture first. Then I'll vibe code to stub out classes and methods, and then vibe code the tests, then vibe code to fill in those stubs, making sure the tests pass. You have to handhold the AI like you would an intern or junior dev.

My car detects an attached carseat and flashes a warning when the car is turned off: please check back seat. More cars should have this warning