In large, powerful and potentially dangerous machines, the firmware may embody safety features. Like preventing dangerous doses of radiation or overspeed conditions on motors or overcharging of batteries.
Why should you have a right to operate the device in a dangerous way that the manufacturer engineered to prevent?
You may be happy to injure yourself and to disclaim your right to warranty but the manufacturer could suffer immense reputational damage if, say, your hacking the firmware of their humanoid robot results in it tearing off your head.
That reminds of this company that used to provide a body implant for people with certain issues. The company ended up closing down and had to go in and remove those implants from those people.
Those people rightfully were like why can’t we keep it and the company was like we have to run the services to manage it. Why not just open up the hardware and let it continue being used? Well if you put the wrong software and it ends up killing the person, the hardware company can be held liable.
Made me realize my dream of having an open-source car is never going to happen.
All the ECU reprogramming "tuners" have been treating cars like open source for a while now. Any old car has a limited amount of SW too, and people have simply switched parts around and hacked them however they see fit.
Basically, we already live a reality where dangerous machines like cars have been hacked and are operated on the street among us.
This is the same argument that was being made that FOSS is less secure because it's open in the 90s or 2000s. Yet the most of critical infra today runs on open source.
Don't you think it'd be safer if these tuners had access to the source code?
The point at which we are performing unnecessary surgery to remove perfectly functional assistive technology from people because of some legal fantasy is the point where we need to seriously ask ourselves whether the legal fantasy is really doing anything useful, or whether it's merely some self sustaining cancer on society.
From the perspective of a researcher: there is a clear 'yes'. Void the warranty, make sure all certification for operarion is invalid.
That is to a certain point. If deemed misuse would be harm against other people or circumvent export control. You may also restrict, who is allowed to do it. However, if software is the only protection against dual-use, tler might have a general problem.
When people get a tool, they can perfectly well hurt themselves with it.
But at the same time, risk profiles between different tools are different, and it is common to have special rules for some of them. Eg. guns already do.
While we might need to stop an unlicensed SW engineer to accidentally or purposely turn your housekeeping robot into a killer robot, the risk of them messing up their fridge is lower. Yes, your food could spoil and kill you, but your fridge could face a power cut and do the same.
So establish which devices have a higher risk and limit changes to those with more expertise — if it means licensed engineers, so be it.
> So establish which devices have a higher risk and limit changes to those with more expertise — if it means licensed engineers, so be it.
I think the main thing to grapple with is the idea that even laypeople who aren't security or SWE experts may want to protect against the highest risk profile possible, even if they don't personally understand what they're giving up. (And, for most of these people, I imagine they're giving up a freedom they might not have ever thought of using in the first place)
Strawman but, If you want to make a device called "securePhone 2027", where not even an evil maid attacker (say, a national security agency who has spied on you over your shoulder for 180 days and has thus seen all of your passwords) can install spyware on your phone, what sort of qualifications do you need to have in order to purchase (or qualifications for a manufacturer to sell) such device?
Part of the issue in arguments like this is in saying that, because the iPhone is the most popular phone in the United States, they should be forced to compromise on security because enough buyers aren't aware of the compromises they're making in order to have the iPhone. So if you still want companies to be able to sell a brick that never runs unsanctioned7 code, you either need
(A) setting an upper bounds (market share?) to how big a company can get before they must compromise on security and allow people to self-compromise via going through multiple hoops to disable security protections (e.g. disabling SIP on a Mac).
(B) means-testing that the customer understands that what they're buying could compromise their security
(C) some key ceremony for allowing said device to be unlocked. I imagine a 1 month waiting period plus needing to show passport and proof-of-non-coercion across multiple nation-states could work? maybe only if they're not apart of a shared intelligence network coallition (like five/nine/fourteen eyes)?
The fact that the iPhone is incredibly resistant to all forms of widespread malware is really good for society at large. Like, the only known widespread malware (outside of web-based scams calling Alert("Apple has compromised your device, call this number")) has been developer-distributed malware due to chinese developers releasing apps via a malicious unsigned Xcode app [0].
I am not saying it is easy, but an open device can still leverage a secure element, sandbox any applications and have granular permissions.
Just like PKI infrastructure, vendors could embed and tie into a TPM similar key infrastructure that would be the only part you couldn't touch (key storage + key validation and signing implementation so it can't be tampered with). I mean, even there the code could be open and only the key material would be secret.
Esp32 is pretty damn powerful for the cost. We have one driving a display and keyboard, speaking and playing synthesize sounds through attached sound chip, driving an ADC over SPI at a high rate and talking Bluetooth to an attached device and talking wifi to the cloud. On a device that costs a few bucks.
I was a mainframe engineer at IBM back in the day (our actual label then was CSR, which some people derisively interpreted as Card Swapping Repairman - which was very far from the truth).
One of the software guys introduced me to VM. Once I got my head around that, he demonstrated running VM itself - within a VM. My head exploded.
Yeah, they would try to how many levels of VM you could run before it slowed down too much to be usable. About 4 or 5 levels back then.
In VM development, we would run the version of VM being developed as a guest and if you wanted to test its guest emulation, you had to run a guest under it. When I had to do this once, I didn't want to deal with the complexity of creating a mainframe os guest like MVS, so I would run CMS as a guest. I would define an extra virtual CPU, which CMS would know nothing about of course, and SIGP (signal processor) it to run code fragments that I wanted to run on a multiprocessor guest.
When you are emulating guest stuff, VM's view of things is completely different than what the guest thinks it is seeing. So of course most new hardware features have to be virtualized if they are to be used by guests and this means recursive VM's as well. So technical discussions could be something else if there were a lot of context ans POV switching going on.
We have frequent power cuts and follow this same philosophy.
Our first line of defence is the solar system's battery. When power goes out, that will keep things running for some time with the big optional loads switched off - e.g. hot tub, hot water cylinder and pool.
If the sun's shining we can run like this for a long time.
In a solar drought, when the solar battery runs low, the next step is to plug the EV in via V2L and use it as a generator to charge the solar battery. EVs have comparatively huge batteries, so at the cost of losing mobility, which may be important in an emergency, this will keep things going for even longer.
One great feature of using an EV is you can drive to a fast charger, charge up in an hour, and transport all that energy back home.
The final fallback is to plug the 2KW petrol generator in to charge the solar battery. This is enough to keep things going forever assuming a petrol supply. A petrol generator is nicely portable and can be loaned to less prepared neighbours who likely have their own petrol.
The key to it all is the solar battery. That gives you uninterrupted power while mucking around with the alternate sources. It doesn't even have to be all that large if you have an EV.
In the scenario mentioned in the GP post where a giant tropical storm hit resetting the grid for 7+ days, how big is the chance that your solar panels installation has survived? Also, if the grid is gone, won’t the fast chargers also be out of service?
Don’t get me wrong, I think you’re right for most cases but I also think for some more end of the world type of scenarios there is little that beats a generator and stored fuel.
If you can stay there (no evacuation for building doesn't seal enough to make it habitable), most of the solar should have survived. Moving a broken panel out of the way and swapping one from the end of the array in to make the cables fit, or even having a spare jumper cable to swap in place of a broken panel, is quite easy to do. Mostly just make sure there's no load current at all on the string when you disconnect the MC4 connector with the special tool that keeps uneducated people from unplugging them under load and getting greeted with a persistent low current (not "arc flash", we're in the region of "high powered welder used with just normal welding PPE") DC arc (that may supply several hundred volts for a long drawn out burn). As long as it's around substantially below 500 mA you're not gonna sustain an arc.
Indeed, that's why I mentioned petrol generator as the last level of backup.
Though the most recent big tropical storm we had a few years ago, the power was out for 3 days, it was dark and cloudy obviously so no solar, the local petrol pumps had no power and the roads were blocked with fallen trees so we couldn't even drive anywhere. In that situation it really is candles and cooking on wood fires once the petrol runs dry.
We're building an esp-32 based hardware golf product.
Esp32 emulation in the browser is an incredible superpower for hardware manufacturers.
You can put your hardware product, with a pixel perfect display and sound, directly into users (virtual) hands so they can try before they buy.
There's no real cost to you, since their web browser and computer provide all the horse power, you just serve it up as a static web page.
In the age of AI, a moat for hardware companies can be network effects through providing an AI friendly base that your users/players can build on.
For us that takes the form of an open source app store where players can extend their device using AI ("Hey Claude, make me a game like this one, except that a giant bowling ball with our company logo embossed on it should drop and roll down an Indiana Jones styled marble run").
It's hard to overstate how powerful it is to have users trying your device, navigating the menus, extending it's functionality themselves with AI, all on their own computer, creating games for others, all without even buying your device.
The really sad thing is that the natural instinct for many will be to screenshot the page and upload it with "please create me a home assistant dashboard in this style".
It's not sad, it's literally what those tools are for.
Yes, this is the "you wouldn't download a car" moment. Yes, I would.
(And no, I'm not doing it, because I have my own preferences, which involves hitting a very small target in design space, that is applying "dark cockpit" principles and a mix of industrial HIGs and nuclear reactor control room interface standards to LCARS.)
He didn't say it's illegal (even if it's debatable and kinda loophole-ish, but lets pretend for now that society has fully embraced a world-encompassing copyright exception if you're named Altman or Amodei).
He said it's sad, and no matter how you want to spin it, it still is incredibly sad and depressing that we're hailing the age of "semantic photocopiers" as some sort of intellectual achievement.
If you consider it incredibly sad and depressing for the majority of people to have cheap access to things that would have cost a small fortune due to being inevitably hand-crafted only a few years ago, I hope you're renouncing clothing manufactured using automated looms, printed books etc. as well?
Absolutely, I think fast fashion brands like Shein and TEMU-style shops are incredibly sad, toxic for the economy and the environment, and I've never bought anything from them.
Are you arguing that the only type of clothing an automated loom can help produce is fast fashion? Is every piece of clothing you own hand-woven, crocheted etc.?
Man I hate this argument. You're comparing cognition to reproduction. You're giving examples of duplication but comparing them to unique artifacts. It's not the same and you must understand this.
On top of this is the gross ethical concerns where all AI companies were allowed to steal all of this information/wisdom/art to train these models. At least with the power looms, a few skilled craftsmen were hired to help design them. This is blatant intellectual theft we're choosing to ignore because we're addicted to cheap products.
We choose temporary sensory pleasure over a healthy society.
I don't think the closest T-Shirt comparison is quite exactly either of those. E.g. you can get any image you like printed on a T-Shirt without hiring someone might be closer, though still not really the same. Thinking about that a bit more, everything seems like unique cognition until the technology to treat it as simple duplication exists. "Duplicate this project with this layout and style" has crossed that line now, just as taking a photo with a filter did long ago which itself long ago replaced a painter and so on. They technologies often produce something of lower quality than what the people they replaced would, few would say a photo filter is as good as a custom painting, but they tend to do the job well enough it replaces most of those kinds of workers.
I think the concern should be more on "you can't just extrapolate the same process forever and expect the same kind of outcome". Just because it was good in small amounts before doesn't necessarily mean it's inherently good to have an infinite capability to do that.
Regardless of all of that, I think it's important if we do have the AIs we're making we treat it as a common good in terms of wealth (regardless if people were paid) rather than a way to funnel a small amount of money to the companies who initially drive it as we are now. I.e. don't even mind if we "steal" (copy) everyone's available outputs as long as we all decide it should really benefit everyone rather than a handful.
> You're giving examples of duplication but comparing them to unique artifacts.
No, I definitely see the difference, but I'm acknowledging that many people probably never cared all that much about how a beautiful thing is made, or at least not as much as they care about being able to have it.
I think it's a net good for the world for more people to be able to afford the things they like (after accounting for all externalities), even though of course it sucks for the craftsmen that lose their economic basis in the process.
> At least with the power looms, a few skilled craftsmen were hired to help design them. This is blatant intellectual theft we're choosing to ignore because we're addicted to cheap products.
I'm pretty sure this was little consolation to e.g. the Swing Rioters [1]. I suspect that this only seems less bad in hindsight. For many people whose livelihood was destroyed by mechanized agriculture, the power loom and other advances of technology during industrialization, I'm pretty sure it seemed equally catastrophic, if not more (18th century government welfare programs were somewhat lacking compared to modern standards).
All that said, while I do believe in human rights, I don't think there should be something like a "right for everyone to continue living like they are used to", which ultimately always means something like economic protectionism, artificial/subsidized demand etc. I'd much rather we find dignified ways to manage the transition for everybody than to engage in some elaborate pretense of economic necessity when there isn't one. If we only grant value or even basic human rights to economically productive workers, maybe that's something worth addressing anyway?
> He said it's sad, and no matter how you want to spin it, it still is incredibly sad and depressing that we're hailing the age of "semantic photocopiers" as some sort of intellectual achievement.
We've been talking about being sad about a lot of different things in this thread.
But here's another thing to be sad about (somewhat related to what I was originally sad about): building "semantic photocopier" is probably quite an intellectual achievement, but so was the atom bomb. intellectual achievements, even impressive ones, can't be assumed to be good.
To hail a "semantic photocopier" requires a pretty myopic and parochial view. Looked at holistically, this stuff amounts to a project to rub out the human and replace it with cheap, inhuman slop. One aspect if that is when I see a picture like this, it's ruined because when I look at it I'm unsure if I'm communing with a human or getting fooled by a simulacrum. Another aspect is how the slop undermines the viability of creative and intellectually engaging professions.
We really ought to look at the kind of world we want to live in (which would probably have maximum satisfying creative work, and a minimum of drudgery), preserve the good parts of what we have, and focus on intellectual achievements to subtract the bad parts. But instead we're just mindlessly building whatever we can manage, damn the consequences or end result (which seems to be: more drudgery, less creative expression).
> It's not sad, it's literally what those tools are for.
If someone spreads poison on your street to kill the strays and mutts (yes, that’s a thing that happens) and it kills your dog, would you like someone to tell you “that’s not sad, that’s literally what poison is for”?
Using a tool for its intended use has zero relation to it being sad.
Yes, I do see that problem. I just wish we'd call it what it is instead of drawing imprecise and confusing analogies that are often thinly veiled appeals to emotion.
That would be the first step to addressing the actual issues: Who will advance the field, assuming that generative AI will always be limited to copying instead of innovating? What will be the place in society for the people that lose their job due to these changes? etc.
> I just wish we'd call it what it is instead of drawing imprecise and confusing analogies that are often thinly veiled appeals to emotion.
And I wish people on HN would steel man arguments and not assume intentions in others, but here we are. There was no “appeal to emotion”, veiled or otherwise. The example used harming dogs because hopefully most¹ of us can agree that’s repulsive and bad.
Everything else you’re saying has nothing to do with the point. AI isn’t the point (hint: that’s why I didn’t mention it), the point is spelled out in plain English at the end of the comment:
> Using a tool for its intended use has zero relation to it being sad.
¹ I promise you that qualifier is necessary. I’ve seen some vile opinions on HN.
My comment was specifically regarding your example of harm against animals, yes.
Why do you need to invoke an image of something widely considered repulsive to strengthen your point against something much more nuanced/controversial?
It was not to “strengthen the point”, it was to have a clear example where most of us can be on the same page of what is wrong. This is HN, anything other than an obvious example attracts an infinite swarm of nitpickers arguing that that whatever bad thing “is very good, actually”.
I designed my basement renovation using a combination of image modelsand a 3D scanned Houzz Pro FloorPlan, and I gotta say it REALLY helped the contractor deliver on the vision that I had.
Decorating when you can easily swap out pieces and parts in images with realistic dimensions makes rapid iteration on physical things much faster and it's easy to figure out what "speaks" to you. Great use of these tools.
I tried a couple of the web-based design-a-floorplan sites and all of them were kinda like using crayons to draw a circuit diagram. Ended up doing a pile of sketches on paper, then when I had what I liked finalised it in Visio and sent it through to an architect who turned it into proper plans which, since they were architect-drawn plans I had to check for f_ckery and then the builder checked it for stuff I'd missed but ended up with a rebuild that was exactly what we wanted.
Well with one bit of f_ckery that both of us missed until after the concrete had been poured because it was too small to see easily on the plan, and a second also too-small bit that I caught once they'd put the framing up and asked them to take it out again.
I’m genuinely curious how you use image models for this type of design. I need a way for my wife and I to visualize a remodel project and don’t know where to start or which tools are best for the job.
dump the plan, make sure it has height etc, after dump pictures if needed,
refs of colors. atropic has some nice recipes somewhere online, entire prompts you can feed into cc. i used their workflows feature, otherwise it's painfully slow.
do it in two steps, first generate just the named 3d model, walls, stairs etc inspect for correctness, after proceed to the fun part, materials, colors, etc.
I did it in extra one step, not sure if that was needed, i build a local harness first for extra small room 6 ft by 6ft, and after that expanded into real size. It installed lots of tools.
It’s not different . One case you chose tiles , you take a picture of your bathroom , and give it a rough sized of tile + room and it will
Give you visual check . Second option if you want it also to calculate materials . This is a 3d path . First one can be done by just chat , another one is coding agent .
yes, it's amazingly/depressingly good at things that would take a $1000+ software, and years of education just a year ago. Now I'm doing this with leftover claude/codex weekly quotas( i do have both due to some testings at work).
Sad, but at the same time it's worth asking how many people would actually pay an artist to do this, versus those who would never do this unless it was for AI.
I want mine in watercolor. I’ve been wanting to take a watercolor class, so I figure this can be my mini-thesis project. I’ll still pass it to an AI to refine and animate.
Sounds sensible. In my country (and probably many others), if you and your neighbour own identical houses or apartments and have identical mortgages, and instead of living in your own one you rent the neighbours one and live in that and vice versa, then you can both offset your mortgage interest against the rental income and both be better off than if you lived in your own house.
I accept the premise on the math, but "better off" is subjective. I'm in the UK so that's my context.
When your renter informs you of the boiler breaking, or a roof needs replacing, that's going to get thorny - "It was fine when I was using it/living in it, did you break it?"
Perhaps the latest round of updates to property law for renters come along and suddenly as a landlord you have to above and beyond what you need to as an owner; you both have to do X, Y or Z to your properties because they're rental and not a basic owner situation.
I have the same memories from many years of working with stored procedures - although you should certainly have been using source control.
But nothing quite matches the satisfaction of someone in a far off engineering group reporting a problem that prevented them from committing data that would have corrupted the database - because your referential integrity constraints protected the database.
In the age of AI and a massive explosion of code touching the database, maybe it makes even more sense to embed logic and constraints inside the database where there's no way to get around them.
Why should you have a right to operate the device in a dangerous way that the manufacturer engineered to prevent?
You may be happy to injure yourself and to disclaim your right to warranty but the manufacturer could suffer immense reputational damage if, say, your hacking the firmware of their humanoid robot results in it tearing off your head.
reply