Hacker Newsnew | past | comments | ask | show | jobs | submit | parsimo2010's commentslogin

In my opinion the separation isn’t between AI vs. not. It’s between software that is restrained and well reasoned in their development vs. software that sticks unnecessary features in without enough testing that creates bugs and pushes useful UI features into a submenu. Software that updates just so they can sell a new version and get a product manager a promotion almost always sucks.

This is conflated right now with AI because every CEO and PM is pushing AI into their product as a first step to increased monetization once they figure out how. This is not well thought out and it ruins the experience of users who were previously happy with their software. A smart company would wait until the AI actually served a real use case and then roll out a well-tested update, but these companies are exceedingly rare.

I honestly don’t care if you add AI features, but make it so I can ignore them and make sure they don’t get in the way of the actual use of your software.

It’s sad that the best example I can think of such restraint is Apple, who was planning on rolling out AI on everything but realized that a lot of it sucked and halted the rollout partway through. It would have been better if they hadn’t made AI the key selling point of the iPhone 16 before failing to deliver, but at least they realized they should stop.

Microsoft also had a partially self-aware moment with their Recall feature, but it took a massive public outcry and their modifications only partially addressed privacy concerns.

But the volume of shit that gets funding because of AI overwhelms even the partial success stories- agents that don’t do what they claim, self driving cars that have never materialized, Rabbit AI, the Humane AI pin, Meta AI glasses, whatever shit Jonny Ive’s deal with OpenAI will produce- all of that makes people very skeptical because product managers and CEOs can’t hold back and wait until an idea is fully baked.


And let me continue my rant: pricing these things suck. You can charge $20-25 monthly for a general AI service like ChatGPT that has a bunch of features like image generation and is very versatile. You can’t cram a shitty feature into your single-application software and charge the same amount. If a PM thinks this is appropriate pricing they are insane.

Additionally, I said previously that I want to be able to ignore the features I don’t use. If you only have one premium tier, and the price went from $5/month to $20+ monthly because you added AI, then I’m probably cancelling because I have to now decide if the features I do actually use are worth the extra money. Plus, companies that do AI this way will inevitably restrict my usage if I do fine the feature useful.

So I guess if I had the choice, I would prefer a company keep a lower premium tier even if I have no access to AI features. But what I would really like is an option to enable whatever AI you’ve developed and bring my own API key, and don’t change your premium tier pricing (or minimally change). But clearly there are enough people out there that are shelling out for the shitty $20+ monthly subscriptions (or getting their company to pay for it) that my opinion doesn’t matter.


Lithium batteries mostly burn when the devices are drawing significant current below the safe cutoff voltage. You can safely discharge a battery to zero chemical energy with a slow draw. They make discharge devices for batteries in RC cars and planes for this. Once the battery has lost enough performance you safely drain it and then dispose of it (not in the landfill because the chemicals are still toxic)- you don't want to leave it partially charged in case it gets punctured in the disposal process.

Letting phone batteries drain naturally is pretty safe, because just leaving it in a low-power mode over time will cause it to self-drain at a pretty slow pace over several months. They should still be disposed of with electronics recycling so that the toxic stuff can be handled, but leaving a phone in a drawer is normally safe. Software bugs that try to turn the phone on into a high-power state right at the safety threshold are the biggest risks, or that you might try to turn it on yourself right at the safety threshold.

Phone fires are typically from software bugs that fail to cut power at the safety threshold, either while a user is trying to squeeze out the last few percent of battery, or if they are in luggage and presumably jostled and buttons keep getting pushed.


Sure, but this is a different value proposition. FB paid $1B for Instagram, which was trendy, growing fast, and already had 30 million users. FB paid $19B for Whatsapp, which was already established worldwide with ~400 million users. These acquisitions were very much in-line with FB's core product. The people saying it was risky were mostly just saying that it was a waste of money and that FB could have just beaten their competition instead of buying them.

And bringing up VR is probably not the best comparison to make- sure, Meta is a leader here, and they are competitive with their AI team too. But "I'm sure it will have huge ROI in the near future" is just saying that it hasn't paid off and they don't have an obvious path to getting there. Shoving VR and the Metaverse into everyone's face hasn't paid off for several years, and the VR segment as a whole has remained niche despite being around for decades.

This acquisition is different- AI is not Meta's core product, it's just something hot right now and CEOs are trying to figure out how to stuff it into their products and hoping they can figure out how to make money later. Plus, they paid a pretty big chunk of money for a company that does, what? Cleans data for LLM training? Meta's Llama team clearly has a good data group already. They paid for a few employees that are clearly popular amongst the executives in the tech industry, but I don't know how this will go in terms of attracting other talent. Unless Wang is bringing something secret along with him, I think this one is an overpayment- Meta will need to both figure out how AI makes them money and Wang will have to attract several billion dollars worth of talent to the team. I'm skeptical that people will talk about this the same way they will about Meta getting Yann LeCun to work for them for a lot less money.


> it was a waste of money and that FB could have just beaten their competition instead of buying them.

Correction, people were saying that FB couldn't beat their competition and had to buy them.


If people were saying that there was no way to beat them, then Instagram and WhatsApp weren't bad acquisitions.


Facebook couldn't beat them on merit because it's not very good at what it does. It does, however, have money because it was good at one thing one time, and therefore it could solve its inability to execute with M&A.


Oh yeah that's bad. I hope there is an option to disable translucency globally. I don't need to see a desktop/home screen under another menu, or even another app under the menu. I can't interact with something underneath the top menu and it really messes with readability from your screenshot.


The way this is usually handled with smaller crimes (DUI) is that the local civilian court gets “dibs” but the military installation can ask to discipline someone under the military system (Uniform Code of Military Justice, UCMJ). Usually the locals are happy to let a military person be disciplined by the military. It keeps the burden off the civilian system, which usually has plenty of other cases to get through. Plus, the military can do things that the civilian court can’t, like reducing a person’s rank.

If the civilian court wants to make an example out of the military member they can opt to keep the case in their court. This can happen if the crime was egregious or there are some other circumstances. Plus, any additional civil suit brought by a victim or their family will always be a civilian lawsuit.

There are times where things are different- in particular, there are times in which something is only a crime in one system but not the other. You can be court-martialled for failing to follow orders, but this is not a civilian crime.

In terms of shooting a civilian, it probably depends on the circumstances. If the Marine was given an order to shoot and had some legitimate feeling reason to do so in the moment, the military would probably do their best to protect the marine, but it would probably be a civilian court trying them (the military won’t take a case if they don’t intend to follow through). Note that for this to be the case, there is probably now an officer who gave an illegal order and the officer would probably be tried for a crime. But there are conceivable ways in which a marine can shoot someone under lawful orders and not really have done anything wrong- self defense is the likely scenario. If a protestor starts shooting a gun toward a marine then they will get return fire.

If the marine were to disregard his orders and shoot someone because he’s trigger happy, then the military is probably going to ask to take the case, throw him in prison for life while demoting him down to E1 (the lowest rank), and generally ruin his life as much as they can. They really crack down on this kind of thing because they rely on discipline to make things work. Marines are generally trained to do as they are told, no matter how much it sucks. And marines that don’t do as they’re told get examples made out of them so that everyone else knows to follow orders.

At least that’s what would have happened in the past, but with the current president who knows how it would turn out. Because the state may choose not to let the case go- the president can pardon a federal/military crime, but not state crimes. So California might keep the case because then the president couldn’t let him off easy.


Andor is shot beautifully but it has a major issue that has plagued an increasing amount of shows recently. It is way too dark. The creators knew it was going straight to streaming and was never shown in theaters. I can’t watch the show in the daytime because even with my blinds down and curtains drawn the window in my living room washes out half of the scenes. Some very important things happened in the dark (it being a spy/espionage show), and I felt like I was blind. The script was not written to be an audio drama, it relies on visuals that I literally couldn’t see half the time.

Directors shooting something for streaming: please watch your show on a laptop or cheaper TV in a realistic bedroom or living room setting (with daylight leaking in or with some lights turned on). We don’t all have reference grade monitors and a pitch black studio. In fact, most consumers don’t have those things. If you really want to keep the cinematic purity, could you at least make a “normie edit” that pumps up the brightness?


I finished season 2 yesterday and was actually thinking how refreshing it was that andor WASNT too dark. It has some dark scenes sure but I really it’s not nearly as bad as most shows these days


Turn on dynamic tone mapping on your tv, or reduce contrast your TV settings.

I’d rather they preserve the dynamic range than succumb to the loudness war.


Agreed. Real PITA with OLED tv’s. Musicians listen to their tracks in car stereos, directors should do what you suggest


Not seen any of these issues on my oled and have watched in full daylight.


I think it depends. For me on a fairly recent OLED, watching from the Disney app in AppleTV it looked pretty spectacular during day and night. I do know _some_ shows are terrible but Andor was totally legible to me. I'm not saying you did not have this problem, just that it's not as bad as in some other shows and personally I could not notice it.


I had the opposite opinion of Andor's cinematography. On a nice OLED, everything looked so gray and flat because most scenes were devoid of true dark blacks or bright whites or vivid colors; like every detail on every scene had to be softly uniformly lit so it could be seen. All the beautiful shot composition was defeated by the color grading and lighting that just screamed that it was targeted towards lower common denominator streaming quality screens and not theaters.

Whole time I thought there was really no point watching on an OLED or in HDR cause it's not taking advantage of either.

You can even see it on the photos in the article. The BTS photographs have contrast and blacks while the stills from the show are muted and gray.

The whole series basically looked like it was trying to recreate the "Shot on Google Pixel" look and completely opposite of HBO's black on black on black.


Netflix has a brightness setting that you can easily get to while watching. I really wish the Disney app had one.


you might have misconfigured HDR in your viewing setup, even if you don't have an HDR display.

a lot of video players don't get it right consistently codec-to-codec, even the gold standard FOSS classics (VLC, MPV) and wrappers like iina on mac.

i typically use iina and vlc as fallback, but wasn't able to get either to play correctly, even though they're fine players for some other examples. i wound up subbing to disney plus for a month to watch it properly.

if you're viewing MKVs of unknown provenance, use an HDR version to ensure it's not a bad encode. if you're not viewing on an HDR display, double-check that tone mapping is enabled and configured correctly.


> Is that a good/bad/ethical/predatory thing?

It is a natural result of our economic system. Economists call it "extracting consumer surplus," and there are several mechanisms companies employ to get the population to pay them the maximum amount of money. Airlines indeed use the information they have about you and the flight you're booking to guess the maximum amount you'll pay- and that's the price they show you.

Obviously we, as consumers, feel taken advantage of because we wish we could pay less (and keep the surplus to ourselves). But this is going to happen in any capitalist system.


This has limited applications. It doesn't have a viable path to being used in a CPU or GPU. So we're not going to see a zillion-fold increase in compute speeds from this. Maybe some physicists find it useful for an experiment, but the average joe won't notice anything different about the world.


I'm curious if this move by TSMC can use this research?: https://spectrum.ieee.org/microled-optical-chiplet

They seem like related attempts at creating optical processors.


OP's link is about a photonic transistor using graphene. Your link is about making interconnects using individual LEDs and fibers in parallel instead of putting multiple wavelengths on one fiber. They are only superficially related.


Small correction: the jammers used are specifically targeting the L-band, because it is used for navigation and satellite communications.

Normally ground transmitters in this band are using just a couple watts or less, so they don’t significantly impact the readings of a satellite looking at a large area on the earth, but a jammer uses a lot more power and can be noticed.


This satellite’s mission is soil readings. Most scientists are not part of the intelligence community. They may have noticed anomalous readings and excluded them from their analysis, but they don’t really have anyone to talk to about the military implications. Plus, while this is cool that you can detect this interference with a science satellite, the major space powers all have military and intelligence satellites that can map the interference at greater precision, so the NASA scientists can pretty much ignore this unless they are particularly interested in the soil readings in this part of the world.


It's not impossible that the Pentagon could have thought "alright, we want these readings. is there a civilian use for this kind of data and decided to see if a civilian project could be sprung up... Though that's more of a Cold War conceit. These days they would just do it themselves, it's probably an easy and cheap project.


Just as there are commercial earth imagery satellites, I would expect there are commercial RF source detection satellites. There are obvious sales channels to hedge funds, countries, militaries, and commercial transmission operators (searching for causes of interference).

Hedge funds is the fun one: detecting economic activity and growth (independent of official government figures).


Thank you for that explanation. It was very helpful. :)


Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

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

Search: