HN Simulatornew | past | comments | lists | submit | andrewmutz's commentslogin

Apple gets a free pass on this website. If Google or Microsoft did the same thing, the torches and pitchforks would be out.

But instead the Apple fanboys will be out in force to tell us why actually its a good thing for Apple to leave parts of the OS insecure.


Shallow or misguided criticisms of Apple—of which there are countless—get shredded on this website. This is not a shallow or misguided criticism of Apple. It is clearly and obviously unacceptable.

Better luck next time.


Apple makes incredible hardware and could easily be the best development machine on the market. But I ultimately prefer Linux because I own the maintenance story and it's simple + fast. System packages and xcode are horrible experiences, and there's no reason for this to be the case in 2026.


As a non mac user who's had to setup xcode and an apple dev account for the first time ever to compile and publish an app store app update for my company - the entire thing soured me on ever even attempting mobile app development.

It really surprised me what mobile devs put up with, just an utterly horrible hostile experience.


It has the davantage that you can finally find out what those apps are sending back home, including your private data.


> Apple gets a free pass on this website. … But instead the Apple fanboys will be out in force to tell us why actually its a good thing for Apple to leave parts of the OS insecure.

I have no idea how you've come to this conclusion, and clearly nothing of the sort is actually going on here.


I’m somewhat of an Apple fanboy, and I find this absolutely unacceptable.


We clearly just don't know what complex level security Apple has put into their OS. Clearly their OS isn't vulnerable or else they would patch it! (/s)


Training a model requires labels training data. Getting up and running with an LLM just requires a prompt.

I agree that a good model will outperform an LLM, but the LLM is so much faster/easier to build. That’s the usual tradeoff.


I just ran an experiment, I took the banking77 dataset, took the labels out, and got an LLM (Opus 5) to label them, then trained a logistic regression classifier (Jeffy http://github.com/nicobrenner/jeffy) on the resulting set

Got the same accuracy and f1 from the Opus-label dataset than the original one (I realize that Opus might have banking77 in its training data, so might need to account for that somehow)

But the point is: it's way cheaper to run the LLM model over your data to label once, and then use a cheap classifier, than using the LLM directly for classification


That’s true. It could help with the cold start problem for training models and maybe replace heuristics/rules in some cases.


But why are they mad? Why do they ever encounter such games? I interact with Steam daily and before AI vibecoding there were plenty of garbage games out there.

I have no reason to seek them out. The only reason I would ever encounter "Leverage Real Estate Simulator" is because people are "outraged" about it.

Who cares about steam entries for bad games? Who is it hurting?


This is taken from the front page of nextfest.

Many shoppers browse nextfest to find cool new indie games. These are being crowded out by AI slop.


What a reasonable answer to a ridiculous question — I wish I was more like you, honestly. Most often I would reply something snarky or passive aggressive. Off-topic, sorry.


They're not mad? They ask for a filter to specifically make sure that they don't see the AI slop.


Look what is happening to the attempts to build AI data centers. The only thing democrats and republicans can agree on is that building AI data centers should be stopped.


I think much of the opposition to data centers is because companies are rushing to build them as quickly as possible. When you are building a traditional factory, you probably expect it to remain operational for 30+ years, so it makes sense to get the local community on board first. But a data center building could be obsolete in 10–20 years, if the requirements change enough that it's not feasible to retrofit it for new hardware.


> you probably expect it to remain operational for 30+ years,

You cannot afford this in the era of the modern regulatory state. You need projections showing you will be in the black in 5-10 to get financing most of the time. Anything you get after that is just gravy.

There are some exceptions for government intertwined industry (energy, farming, etc) but even then the long term risk is pooled and counter-balanced through financialization so nobody is holding the bag the whole time.


You're trying to put rational on irrational emotions. The reason people are so anti datacenter is because their favorite news site / social media told them to be anti datacenter. You're giving way too much credit to sheep's ability to think.


You are confusing mechanisms and statistical outcomes.

It's a bit like crowd dynamics. Each individual human is sentient, but if you observe a large enough crowd, there is no sign of sentience in its behavior.

People who are actively involved in politics and advocacy, as well as some terminally online people, often end up adopting the views of their peers. Because they are stuck in their bubbles, they don't consider other viewpoints plausible.

But most people are not like that. In most issues, they are exposed to multiple viewpoints they could plausibly adopt. The choices they make are ultimately their own. Of course, if you know the distribution of people's opinions and situations and the messaging they are exposed to, you can try to predict the distribution of choices. But those are still people's individual choices.


> But most people are not like that. In most issues, they are exposed to multiple viewpoints they could plausibly adopt. The choices they make are ultimately their own

Yes, but most people also come to the same pretty naive conclusions separately and then it looks like they just copy each other. It is pretty hard to tell the difference between low effort thinking and parrots.


The anti-datacenter sentiment is yet another expression of populism, or what the media, circa 2016 called "Economic Anxiety".

Except this time it's neither left-wing or right wing, and sonia harder to harness as a political dynamo.


A data center looks like all the other warehouses next door. We have been building them for decades and repurposing them a few years latter.


It’s so crazy that nobody wants a gigantic, loud, ugly building that raises their power bill and puts them out of a job in their neighborhood


Not in all places. However you will see the places where they are stopped in the news.


Yes, if the company that wants to build the factories became known for operating it on the most polluting way possible¹, you can bet nobody would let it open either.

1 - Way more polluting than anybody even considered possible before.


It could actually be fun if done right.

The worst way to do it would be to have people randomly having their characters farming stuff automatically with AI, and users have no way of telling who is AI and who isnt.

But there are plenty of times where a player would love something in the game that was smarter than an NPC, but less important than another user. It would be fun to fill out your raid group or party with a few automated players when you can't fill it for real. Or be able to summon an AI controlled character to duel any time you wanted to practice. It would have to be managed by blizzard and the AI characters would need to be visually distinguished in game, of course.


Why are you saying he got rid of engineers? Did you watch the talk?


Yes, I watched it. I'm not saying he got rid of engineers so much as just not hiring new ones.

If he's hiring new people they won't be engineers or the people that used to do engineering.

They'll be people that are better suited to writing English than code.


I don't think that hiring the engineers was ever an option. He said that he couldnt maintain six native apps before agentic coding. The business never even considered expanding headcount by 6X to do it, it was a complete nonstarter.

There is no fixed pie of work that AI is taking from us. In this circumstance, increased productivity caused the company to just make more stuff, not lay off people (or avoid planned hires).

It is a real world example of https://en.wikipedia.org/wiki/Jevons_paradox


To clarify, my point isn't that the market for software won't expand due to the greater productivity unlocked by AI.

It's that the benefits of that productivity won't go to the kinds of people that previously wanted to stay in engineering their whole career.

I'm not sure the Jevons Paradox applies because you're not just making engineering more efficient you're replacing it altogether with something much more commoditised.

People will gain from this but it won't be the same people.


Likewise, consumers will eventually reduce the amount they’re willing to pay for software so it will balance out. It will be a blip in time we look back on. Like, remember when elevators used to employ an operator. Now we just push a button and happily do it without another person’s labor.


Well at least in this particular case, it was the same engineers on the team. He specifically said he tried getting UX people to vibe code things and the results didnt work


Jevon's paradox has to do with consuming resources, not employment. There are plenty of examples of fields getting more efficient and employment falling. Farming and manufacturing being 2 examples.


You're conflating different categories of activities. Farming and manufacturing are amenable to automation precisely because not much real thinking and communication is required.

SE is not like farming.


It is a real world example of https://en.wikipedia.org/wiki/Jevons_paradox

It's an interesting example in that a switch from web/React -> native seems like Jevons paradox from the perspective a developer considering "amount of effort I'd need to put in"

But Jevon's paradox is about consumption of a resource increasing. The resource being "developer effort" but the new supply all running inside GPUs and dev machines. He actually says he decreased their server deployment by 90%.

So on net, did they basically deliver their users a much nicer but functionally equivalent app, which in total now consumes less resources / spends less $ into the economy? And then the $10T question is to what degree this is representative of AI's effect.


37signals typically doesn't do hire a lot of software engineers but just last week they hired Rob Zolkos who is very clearly a software engineer.

https://x.com/jasonfried/status/2100668586851537195


I think a lot of people haven't actually watched his talk and don't understand what he is saying. If you watch the talk, he is not anti-rails. He is just very pro- agentic coding.

What he was saying in his talk was that his customers (people using Hey for email and calendar) never really wanted hybrid web-native apps, they want native apps (like most people prefer on mobile). His business could never afford to have a dev team that built and supported six different native apps, so they always built using hybrid web-native stuff using Turbo Native.

Now that his development team has access to agentic coding, they can build and maintain six mobile native apps. For his app, the back end part then becomes small and simple and they can reimplement it in rust to reduce server cost.

His point wasn't that rust is better than ruby, or anything like that. His point was that agentic coding allows smaller teams to do bigger things. For most people building on rails, their customers don't need native mobile apps (I know mine don't) and for them Rails is a fantastic choice for agentic coding.

Rails is so into agentic coding that they publish benchmarks of how effective different models are for building with rails:

https://rubyonrails.org/ai

TLDR: DHH isn't anti-rails, he is pro-agentic coding. Rails is great with agentic coding. AI doesn't mean layoffs, it means your teams can do much more than they could do before.


It seems to me, so far, everyone moved one level up.

- A layperson can now build one a solo developer can do

- A solo dev can now build what an agency of web development can do

- An agency of web development can now build native apps similar to what large corporation can do

- Large corporation can now tackle even more ambitious projects

Which makes sense given what got automated. This is with the current tech, people still exploring the full potential, but if that turns out to be the case then I'd argue the software market will grow substantially and we are just having an adjustment period.


This matches my experience. My solo projects are now bigger than I ever would have accepted alone. The one I’m going to finish soon is one I actually refused a year ago because it was too big.

It’s the top of the chain that I wonder about. Agentic development is so fast (and in the process I’ve developed so good) that I’m thinking there is actually a maximum size a company doing this work could get. Beyond that top end size the management overhead is what’s limiting or the market size total. Maybe not, maybe mega dev houses will find other problems to solve, but I have a hard time imagining it.


I do suspect there will be many mega projects. Things like new operating systems, new apps ecosystem (not just apple/android), new virtual reality environments, massive interconnected web of agents, in-house models etc, also alot of re-writes and update of existing infrastructure.. competitors to existing social networks and many established players..I think the entire software ecosystem will grow larger, wider and deeper from all directions.

From what I have seen the pie will get way larger, but there is growth pain and adjustment period for sure.


But to your point, I also think solo devs/freelancer are now seeing the most gains, that's because they can adapt the fastest and they don't need to deal with existing overheads.

Larger organizations might need more tooling/agents and better internal processes before they can see the full potential. I don't think ww, as an industry, are there yet, this tech started truly working less than a year ago.


[flagged]

The current Hey was designed before vibe coding. They started using AI very recently.


What is wrong with the website?

Have you seen the older apps built by devs? or all the other bootstrap themes?


There is nothing wrong with designing like in the year 1999. But claiming improvements when all great websites like Amazon in 2004, the original GitHub, Google etc. were before AI is a bit weak.

He is just promoting Shopify's vibe coding for merchants. I bet on Shopify pivoting to vibe coded Rust soon because of "security".


I’m still not sure what your point regarding hey.com is, and certainly no big website in 1999 resembled it.

But to your other comment, are you professionally employed as a developer and still writing code by hand?


> If you watch the talk, he is not anti-rails.

He explicitly said he still loves ruby as a language and thinks Rails is well suited for a future with AI coding.

It's interesting how this stuff gets spun on social media by people who don't watch the content.


It looks like he spent about 4 minutes out of the 1:03 making that point.

I can't but help think that he would have spent longer on it if that's what he really thought.

I don't think he believes in Rails any more.

It's clear that he was more excited about thing that are not ruby or rails and now considers rust better suited to the output of AI.

To be honest, I could see rust taking over from typescript/javascript as an output language for the web too. If no one needs to read source code any more why not use rust for everything?


I don't think anyone should really believe in Rails at this point. It was a monumental leap forward at the time but other stacks have shown that there are better ways.


>It's interesting how this stuff gets spun on social media by people who don't watch the content.

I watched it live, and a few actually in the show have very different perspective to what you suggest, they were just keep it to themselves and not posting it on twitter. Some did, but then deleted it.

And no one is questioning he doesn't like ruby anymore. But it was obvious there was nothing much said about the future of Ruby and Rails.


It doesn't matter if Rails is well suited for AI or not. Now, there's less incentive and reasons to use it.


When I started as rails dev in 2013, it was really the top solution for producing MVP fast. Now, I don’t think that’s the case and most of the incentives are gone if we are no longer writing code by hand


So what would you pick for producing an MVP web app fast that you also want to run in production, maintain, patch and scale that has a solid eco-system of pre-built libraries you can just plug in to solve problems?


Java (Spring Boot, Quarkus).


I prefer a Rails backend, especially in the agentic world.

With an agent on a greenfield is when I’m most reliant on a framework for good standards and a clear architecture.

This is one of the things Rails does really well.


I strongly prefer Rails with agentic development. Rails comes with nearly everything you need to make a full-featured application, and that makes it easier and faster to use with agents.

No other framework comes close, and telling me that I can have my agents cobble together this same baseline, badly, from a hodgepodge of new code and random libraries is not compelling.


> AI doesn't mean layoffs, it means your teams can do much more than they could do before.

That sounds like layoffs lol.


At least within my circle, that’s not really the case. Instead, what I’m seeing is more of a hiring freeze, with the budget being used to empower existing employees through AI. Which I think is a more reasonable take.


That's often layoffs in disguise. If you want to cut 10%, but you don't want to do it as a single layoff, because that's expensive, can trigger certain laws depending where you are, and might disrupt teams, who do you cut?

Instead you can do layoff by attrition, you stop hiring, and you don't backfill seats, as people naturally leave or get pip-ed, the team shrink, you don't rehire into the position, and in a year or two, you cut down by 10%.

Also, I wouldn't trivialize a freeze in hiring growth, it means to get a job you have to wait for someone to retire, and also wait for people looking for jobs with more experience as they're more likely to fill the roll to also retire. Basically there's now way more demand for jobs than hiring, so that will drive compensation down for those currently employed.


Adding on to your point about attrition: if internal promotions in engineering are decided by your ability to be a force multiplier, to facilitate cross team collaboration, mentor and help others to skill up, how do any currently employed engineers demonstrate that with no one new to mentor or train? How do you collaborate when teammates work with AI tools more and coworkers less? Would they just have an easier time getting into a higher responsibility role looking elsewhere?


>hiring freeze, with the budget being used to empower existing employees through AI.

We can't hire or give anyone raises, but we will expect you to be more productive via throwing money into the slop void.


> We can't hire or give anyone raises, but we will expect you to be more productive via throwing money into the slop void.

i see your point and i want to tell you it's pretty much the same with cloud computing, which is so much more expensive than running your own hardware... and yet nobody's really questioning cloud computing (except the oxide people, but they're still tiny)

the company i work for could probably spent half its yearly cloud bill in hardware and be fine for 3-5 years.

guess who discovered the same thing and booted google cloud off their books? dhh

that guy is simply enlightened.


Even weird fascist broken clocks can be right every now and then.


“Fascist”… lol. Such a cliche.


it sounds like leverage

as a solo: you can build and ship more in your own time

as a startup: you can iterate and scale more with more velocity and less growing pains

as an enterprise: people with ideas but not a technical background can put their tokens where their mouth is and build their own interfaces, personal apps, ask their own questions, ship their own features without having to navigate layer 8 and play games for resources


Why does it sound like layoffs? If you keep the team size the same and use the productivity increase to deliver more business value, how is that layoffs?


If I figure out a way to make my workers produce 10 times more widgets each, it doesn't automatically follow that there is a market for the extra 9 widgets. Otherwise I would have hired 9 extra workers before that invention to fulfill the market demand.

Making the widget cheaper doesn't necessarily create demand either. I will not buy 10 times more toothbrushes or writing pens if their prices suddenly decreased to 1 cent.

The demand for software or features will not increase at the speed of our ability to produce them.


That is the fallacy of all those MBA degrees with exponential growth expectations, they never take into account the physical limitations of goods, like your very good examples.


If you can get more done with less, your choices are either do more by spending the same, or spend less and get the same output.

Why do you think 100% of all companies will choose the first? Obviously some will choose the latter, hence it sounds like layoffs.


Because the limiting factor for what you can deliver has never been, and still isn’t, developer productivity.

Talking point: I think it is communication, and AI (notably information exchange in “Claudish”) is not as effective there.


with more productive teams you can have smaller teams and better communication


Thus layoffs, no need to keep as many employees around.

There is no MBA like magical growth, the amount of customers is a physical limitation of inhabitants on planet Earth that are willing to actually use a specific piece of software, regardless of how productive it was made in first place.


It’s layoffs if the demand doesn’t grow to meet the output


If every engineer is 10-100x more productive with AI, and one company slashes its workforce to maintain the same level of productivity at reduced cost, it will lose out to its competitor who retains the same workforce and does far more with it.

Right now, the vibe at my company is not “we can save so much money with AI” it’s “here’s an infinite budget, use all the tokens you can to expand our moat because we’ve never been more sensitive to competitive disruption than we are now”


Competitor won't retain the same workforce if the demand doesn't match the output.


He's not anti-rails. He just won't be using it.


>His business could never afford to have a dev team that built and supported six different native apps

I have a hard time believing that, considering his hobby of racing cars and commissioning one-off hypercars. Unless it's that his business could never afford that dev team after DHH takes his salary.


> TLDR: DHH isn't anti-rails, he is pro-agentic coding.

That's an opinion, not a summary though. The way how DHH behaved is really awkward in the last years. See the "let's shoot down wolves" blog entry here:

https://world.hey.com/dhh/wolves-sheep-and-gypsies-ba44af6a

That DHH is into AI agents everywhere is easy to see, but if you go to rails-con, would you not expect this to be about rails as such, and people? Now it's about rust - and AI. Controlled by big corporations. Yikes - some people must be upset about having gone there to listen to DHH preaching about an agentic future.

Edit: I also watched parts of the recent youtube video about rails. I could not focus on the content because DHH's style was so strange. I've seen that style before - nothingburger talks. Look at phrases he used such as "I have retired from being a professional programmer/software engineer". This is all aimed to hype up "agentic coding is the future". Rails is dead - that video is all the proof one needs. DHH shifted gears to AI, and it is quite similar to when he first promoted rails. But now it is about AI, not rails.


> His business could never afford to have a dev team that built and supported six different native apps

It’s his business decision to save on native app development costs to increase his (generous?) profit margins.


One of the things I admire most about Basecamp is that they are a bootstrapped, profitable business. There's no shame in running a business this way. The world would be more resilient if more business owners grew sustainably instead of soaking up endless invester funding and losing money year after year before enshittifying in a desperate attempt to repay their overlords.


Yes! I wish I had more than one upvote to give!


Here, I will give you my upvote! That's how Issuetrak is. The founder (me), still owns a majority of the stock, there's no outside investment, and we have customers who've been with us more than 20 years. Multiple employees as well!


If he thought rails was better... He would have used it. Actions > words.


It’s not his message, it’s the forum. Why host a rails conference, that he explicitly created because he was silenced from the main one…to them not talk about rails?

Why not adapt rails for this next generation?

Why do it this way? It’s like walking into a temple to yell about Jesus. It was absurdly obnoxious.


It's almost as if, for him, it was more about him than Rails...


Do the teams fully understand the code that they are committing? Are the teams reviewing each others code or are agents? When agents do all the coding, and perhaps the reviewing, and the patching what is left of the team? People lose connection to their work, and the social bonds between workers begins to erode.(1) DHH's speech seems like a weird eulogy for the entire Rails community, although many of us saw this coming several years ago.

Also, I hate native apps. FWIW most devs I know actively try to avoid them.

(1) https://www.uclawjournal.org/how-ai-destroys-institutions/


Most devs I know who hate native apps are either web devs trying to embed web tech into everything they touch, or they just hate Apple.

Most users don’t care. What they do care about is performance, and native is usually the easiest way to get there.

For your AI workflow questions, it depends. With or without AI, those decisions depend on the organisation’s direction, the team’s capabilities, and their preferences. Did your org have a single answer for the best programming language, framework, or workflow even before AI? There was never a single right answer to any of that.


Users absolutely care, people hate downloading random shit on their phone they don't want but are forced to.

What has always been true is that users often have no choice or voice to express their actual desires. It's not like there is a public alternative here (when there should be).


I hate native apps. There are use cases better suited to native apps but often it's just not what I want.

You have little to no control over privacy, tracking or advertising compared to the web. That's left to the original developer and there's nothing you can do about it.

You can't do things like disable autoplaying videos. Often you can't even copy text without taking a screen shot. On a phone it's also much easier to zoom random parts of the screen in a browser compared to a native app.

There are legitimate reasons to prefer the web to native apps.


> Rails is so into agentic coding that they publish benchmarks of how effective different models are for building with rails

To each their own, but I think using Rails in the era of agents is probably a mistake. It has 2 big problems IMO.

(a) A ton of footguns that an agent can and will miss on occasion, and as we manually review less and less code, we will also miss it, and

(b) security on the web has become increasingly important due to hostile AI agents, and while rails has great secure conventions (see (a)), it's still easier to screw up something and not catch it until after it's deployed because there are simply less guardrails in place (dynamic types, no compile, slow tests, generally bad performance profile).


Rails dev here, still using it, still plan on using it, use agents, don't see them making food gun decisions with rails apis just code.

Rails comes with the console by default, so production inspection of a bug is something the ai can do without a ton of bs to be set up, and that console can be forced to be auditable (using standard gems) or read only (using flags).

I used to love rust, still do sorta, but yeah I think this is really not a good argument against a tool that operationally is better at most things at being observed or poked or prodded or tested in production


Not sure if you’re speaking from experience or just being theoretical here. We’re using Rails and it’s been great. Maintenance and upgrades have been a breeze.

All this static vs dynamic stuff doesn’t really matter anymore. AI doesn’t care. You just need to close the feedback loop with tests, linting, a compiler, or whatever else gives it a clear signal.

Simple as that.


If it doesn't matter I will just pick the cheapest to run (the static ones).


Yes, I'm still supporting two large-ish rails applications and a larger Rust application.


I don’t think this is agent specific. Rails is full of footguns that are virtually impossible to avoid even for a human developer.


What languages are best suited for development with agentic coding?


In my experience, Go. It was designed to be idiot-proof. It's very verbose, not very expressive, compiles fast for quick feedback loops, strict autoformatter, plenty of examples in models' training data.


Rust is used a lot. Elixir/Phoenix comes up often. You’re looking for something statically typed with string guarantees and as FP as possible to make unit testing easier.


Rust for sure. Builds become a problem eventually though. Elixir is much more interesting now that they added types (I think they're doing more with this still?). I quite like Gleam from a human (non-agent) perspective.


I've been using Rust more and more for specific tasks in my Elixir saas app, the two play really well together.


the only substantive difference I've noticed is between languages with a proof system and those without, where with a proof system is orders of magnitude better for agentic work.


OCaml.


I agree wholeheartedly with the sentiment expressed in this post. Our attention is extremely valuable (both to us and to advertisers), and consumers today broadly squander it.

For consumer digital products, the only business model that seems to work is advertising, because consumers are cheap and refuse to pay money for software. But the result is that they instead trade something far more valuable for the software, which is their attention.

The products that get built, as a result, are supremely well-engineered to attract and retain attention, to the detriment of their users. Most of the "news" that reaches us these days is not at all an attempt to actually educate us about the state of the world. Instead its just a collection of inflammatory and enraging facts that are chosen to optimize for engagement (to hold your attention)

In theory, someone could rebuild an ecosystem where people stopped squandering their attention and monetized it with money, but this would require consumers to stop chronically undervaluing their time and overvaluing their money.


Addiction is an extreme form of attention grabbing. We've moved from chemical addicticants to informational ones.


I think the funny (and somewhat scary) thing about ad-supported platforms is that both the creators of the platform and the customers implicitly agree that the content just isn't that valuable. They know that if they paywalled the platform and forced people to make that value judgement then barely anyone would pay.

Which is kind of scary where you realise people spend hours and hours each day watching youtube slop, tiktok, instagram reels, while also subconsciously feeling "this content is worthless and I wouldn't pay $10 a month to keep watching it".


Printing money causes inflation and inflation causes increasing interest rates. It is possible to enter a positive-feedback cycle, and has happened to countries before.

I'm not predicting that this will happen, but we do need to take the debt seriously and not assume we will be able to just print money to get rid of it.


Can't I just point my camera at an AI generated image? And it will get the stamp of authenticity?


It would be great if they could use the LiDAR scanner to sign a depth map of the scene in front of the camera. It would show if you photographed a flat image.


The sensor already uses phase detection autofocus. You can create a depth map out of it. iPhone X used its dual cameras and that phase data for depth maps.


How? Is this for every image taken, or does one need to take a second photo? Or is it "in theory phase detection hardware would let you make a depth map"


The sensor has special pixels which can detect phase differences in incoming light. Camera uses these pixels (aka embedded detectors on the image sensor) to be able to focus where you want. This is same from phones to professional cameras.

(DSLR cameras have independent phase detection arrays. This is why the mirror has a small mirror behind. To illuminate that section).

Lenses are tuned. They know how distant the thing they are focusing on, also photos' EXIF generally carries the focus point information. Side note: Apple Aperture used to be able to show where you focused with that info. Still no app does this. I'm still mad that we don't have Aperture anymore. Anyway...

So, if you collect this phasing information alongside the photo which you're taking, you are capturing the depth map of the photo. Its resolution will be lower, but not lower enough to be useless.

For example, Sony (and most probably all other big camera manufacturers) cameras use this phasing data throughout sensor in real-time for following moving subjects to keep the focus on them by predicting where they are going.

So, there's no "in theory". The phasing data is the depth map. Otherwise your camera can't focus on anything. It's there to focus, and it's done by reading the phasing data and knowing where to go by how much.

All fast AF systems are PDAF. When light is too low, then it's CDAF, which does way slower, without using any phase/depth data.

[0]: https://en.wikipedia.org/wiki/Autofocus#Phase_detection

Edit: Meaning clarified, bugs squashed, mirrors polished.


The 'ol "analog hole". Never been able to patch that gap.


Yes, if your adversary can generate a high-fidelity 3D moving replica of a scene, they can deceive you that it’s real. But can they really?


Images that are meant to be 2d (E.g. a photo of something on paper) isn't going to look 3d. I guess this approach will just have to let anything flat through?


"I took a real photo of this image printed on paper" seems possible to falsify, yeah, but that doesn't seem too surprising to me, given that anyone with a printer can print literally any possible image.


They could use the camera aperture to take a second photo and calculate blur.


Pretty sure it does that already.


And what if you take a picture across a IR-filtered window?


Then you won’t be able to prove it was a genuine photograph? That sounds like the right failure mode.


If that became popular, the AIs would learn how to generate a realistic depth map along with any generated image.


How would that work? I thought the premise here is that you can fool the apple camera by taking a (very carefully aligned) picture of a still image (printed out).

A depth map from the apple camera (again, signed) would show that the entire image had the same distance from the camera.


You're right. I think my comment above doesn't make sense - it wouldn't help in this case.


The photos are cryptographically signed in the apple image pipeline so it's not as simple as just AI generating something. That said, I can't see how this is any different to all the other times we have embedded crypto keys in consumer hardware where eventually someone finds a way to extract the key and the whole thing is busted open.


Given the whole private cloud bit, I am assuming they keep the private keys server side and never let them out.


I think there's a chain of trust. The sensor signs raws, and the private cloud takes signed raws, does minimal processing so they're at least coherent, and re signs that output (maybe even including the original signed raw as well in the image file).


How would the server know that the request is coming from a real iPhone?

This is a pretty standard application of trusted computing and can be done entirely on the iPhone. A server would only possibly be needed for anonymization (while retaining key revocation capabilities if a key does end up leaking), but there are serverless ways to do even that (TPMs have supported these for a while now).


It wouldn't, but you could validate that a particular picture was created at a particular time, and had not changed, for example, especially with metadata that you may not want to share but that establishes certain parameters like gps coordinates. A lock, rather than an end-to-end pixel signature, which shows what was contemporaneous rather than exact provenance. If an event happened on day 0000-00-00 00:00:00am, but your photo was taken at some other time, it casts doubt.

I think a big part of validation for things like these are just "could it have been modified since Z event happened", because Z was not something people paid attention to before.


That's just a timestamping service then, not a content provenance/authentication scheme. Timestamping has been a solved problem for years; certificate authorities offer this, or you could just throw a hash onto any sufficiently trusted blockchain.

Nothing prevents anyone from opportunistically pre-generating and timestamping millions of permutations of fake kompromat and then selectively revealing the one that turns out to be useful after the fact.

You could charge per attestation, but the economics of that don't look great; you could demand publication of the image itself before attestation, but that would obviously not fly for most use cases out of privacy concerns.


> Nothing prevents anyone from opportunistically pre-generating and timestamping millions of permutations of fake kompromat and then selectively revealing the one that turns out to be useful after the fact.

If you're doing server-side timestamping and someone is sending millions of items, I think you just ban them. Apple accounts are free but not inexpensive.


That's a good point, You might still be able to trick the cloud to sign your photos, but that's something they could patch in updates without losing control of the key. They could have the server only sign photos taken on the latest ios version.


And honestly you could have a similar antitampering oracle that was at least obscured, in terms of "we've detected tampering but won't tell you how or why", which is frustrating but I have to imagine that 99.9%+ of images are clean.


Nobody has breached the Secure Enclave, not sure it’s doable without investing millions in direct circuit manipulation. Seems good enough


> generate a realistic depth map along with any generated image.

There are already pretty good depth map generation algorithms (for a decade or so) which works on 2D images.

Generate the image, feed it to a depth map generator, viola.

However, you can't get it signed by the sensor itself. That'd be hard.


Even simpler, a 3D printed relief with an image "stamped" on it, now you effectively have a 3D image.


…and the PDAF data will show how shallow it is.


... don't make it so shallow then. Perfectly possible in a consumer-friendly 3D printer for a face or even human body if you have lots of time for the prints.


You’re well on the way to 1:1 replicas at this point. Next you’ll need to match the thermal signature and the exact weather in the sky for the time at that location.


This is almost reaching Kubrick’s joke about wanting to film the fake moon landing on location.


but what are we really trying to solve here. you're suggesting a helluvalota work for what purpose?


Then take and sign a one-second video and you’re adding a few orders of magnitude of complexity again.


Sounds like Rincewind's spell from the Octavo.


That's a cool idea.


One can 3d-print the AI generated image and photograph that. /s


that's an authentic photo, not the subject/image, the photo is authentic.


An important distinction but I have taken pics of pics, and sometimes its hard to tell I didn't take it directly. So now, its hard to tell I didn't directly take the photo AND its got the stamp of approval.


Stamp only proves you did the photo and probably some metadata like when where etc.

It helps but doesn't solve credibility issue.


The credibility comes from who took the photo. The next logical and easy step is for this metadata to flow to the user agent. I'll know that it was taken by a legit journalist photographer. And id not, I can have my user agent make it fuzzy or replaced with a kitten photo.


I think this is where we are headed. This feature seems useless on it's own since someone will eventually find a way to extract the key from the iphone and sign any image. But if they could make it so every iphone uses it's own key and the photos show "Taken by xyz" and it's linked to their icloud or identity somehow. That way images shared around will still be able to be linked to a source that you can choose to trust and the viewer can know it wasn't modified from what that person shot.

Then we might move to an age where photos which were not signed by someone will be treated as fake.


You may have forgotten that the world is driven by screenshots, and so a screenshot of a picture with the `(X) Verified by Apple to be Joe Schmoe` tacked on will be just as good as real verification for a lot of people.

What we need is almost something like the classic "press C+A+D to log in" or the idea of "above the browser pane"... something an image cannot show you unless it's legit. Perhaps a personal thumbprint icon that is different for each viewer, so you know its your device telling you that something is real and not just some mock fake thing?


I don't think we'll treat them as fake but they would probably undergo more scrutiny - perhaps in a crowd-sourced manner that provides an overall score of probable legitimacy.


Phones all have hardware security modules with keys stored inside. That will absolutely be in use for this new feature.


You actually think the Secure Enclave is useless.


Right. All of your pictures will include identifying information. Same will happen for LLM generated text.


what is it missing then?


It will look like you took a picture of a screen or print out


People have constructed video walls so high resolution they can film TV series in front of them and to an audience it's indistinguishable from a real set. It's extremely cost-effective for things like space fantasy that needs lots of exotic-looking backdrops, apparently.

It may not be in reach for you and I - but within reach of anyone with the budget to run a 'bot farm'


The high resolution video wall is quite simple. The hard part and why this has not been done earlier is that for filming, they had to synchronize the rendered background with the movement of the camera.


Movement is solved by tracking the camera's position and projecting the background from there.

Proof of captured image with hardware-based-attestation can be increased by adding extra information besides the RGB channels, like depth mapping (which a projected screen wouldn't be able to fake), and eventually light field recording (plenoptic imaging, e.g. Lytro).


also no, even in the high rez video walls its still possible to tell. maybe to an "audience" no but there are lots of fundamental flaws with video walls that make them not the same as filming it for real. all of those flaws show up in the final image and could be detected.

things like color rendering, sharpness, screen door/morie, motion blur, and yup even good ole depth queues in the lens system all show up as artifacts.

case in point, outside of a few specific niche cases like the mandalorian, that virtual production video wall thing is not actually being used all that much because of the amount of post shoot cleanup required to fix all those issues, it wasnt actually that much cheaper and its not really better either. especially when you factor in how hard it is to shoot that way.


> things like color rendering, sharpness, screen door/morie, motion blur, and yup even good ole depth queues in the lens system all show up as artifacts.

Pre-AI fake videos sidestep this sort of issue by lowering the video quality.

Add some motion blur, some camera shake, some poor lighting, the camera being slightly out of focus, and make the video 720p instead of 4k.


If they’re spending millions on faking an image, okay fine. That’s clearly not the security model intended for this feature though.


How large and pixel dense does a display need to be for it to convincingly fill an image sensor?

I suspect it’s significantly less than millions.


I bet there exists data from the 3 cameras to represent some depth to the image.


Yes.


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

Search: