orclev

orclev , 5 days ago

I feel like Google is trying to speedrun killing their search service. It was already significantly shittier lately just because of all the sponsored bullshit and SEO spam, but this rush to cash in on the AI fad has really destroyed the last utility it had.

orclev , 5 days ago

The implication of a 200 to 1 algorithm would be that the data they're collecting is almost entirely noise. Specifically that 99.5% of all the data is noise. In theory if they had sufficient processing in the implant they could filter the data down before transmission thus reducing the bandwidth usage by 99.5%. It seems like it would be fairly trivial to prove that any such 200 to 1 compression algorithm would be indistinguishable in function from a noise filter on the raw data.

It's not quite the same situation, but this should show some of the issues with this: https://matt.might.net/articles/why-infinite-or-guaranteed-file-compression-is-impossible/

orclev , 5 days ago

GPLv3, make it really radioactive to them.

orclev , 5 days ago

They did, but then Elon insisted they add a virtual neuralink into it and now the neural network is braindead.

orclev , 4 days ago

It's important to distinguish between lossy and lossless algorithms. What was specifically requested in this case is a lossless algorithm which means that you must be able to perfectly reassemble the original input given only the compressed output. It must be an exact match, not a close match, but absolutely identical.

Lossless algorithms rely generally on two tricks. The first is removing common data. If for instance some format always includes some set of bytes in the same location you can remove them from the compressed data and rely on the decompression algorithm to know it needs to reinsert them. From a signal theory perspective those bytes represent noise as they don't convey meaningful data (they're not signal in other words).

The second trick is substituting shorter sequences for common longer ones. For instance if you can identify many long sequences of data that occur in multiple places you can create a lookup index and replace each of those long sequences with the shorter index key. The catch is that you obviously can't do this with every possible sequence of bytes unless the data is highly regular and you can use a standardized index that doesn't need to be included in the compressed data. Depending on how poorly you do in selecting the sequences to add to your index, or how unpredictable the data to be compressed is you can even end up taking up more space than the original once you account for the extra storage of the index.

From a theory perspective everything is classified as either signal or noise. Signal has meaning and is highly resistant to compression. Noise does not convey meaning and is typically easy to compress (because you can often just throw it away, either because you can recreate it from nothing as in the case of boilerplate byte sequences, or because it's redundant data that can be reconstructed from compressed signal).

Take for instance a worst case scenario for compression, a long sequence of random uniformly distributed bytes (perhaps as a one time pad). There's no boilerplate to remove, and no redundant data to remove, there is in effect no noise in the data only signal. Your only options for compression would be to construct a lookup index, but if the data is highly uniform it's likely there are no long sequences of repeated bytes. It's highly likely that you can create no index that would save any significant amount of space. This is in effect nearly impossible to compress.

Modern compression relies on the fact that most data formats are in fact highly predictable with lots of trimmable noise by way of redundant boilerplate, and common often repeated sequences, or in the case of lossy encodings even signal that can be discarded in favor of approximations that are largely indistinguishable from the original.

orclev , 9 days ago

Unfortunately the fact that NFC can't be used on anything that's rooted anymore is kind of a deal breaker. If I could use google pay and my normal banking apps with GrapheneOS I would switch to it today.

orclev , 9 days ago

Ultimately the real solution to a lot of these problems is likely to be a Linux phone OS. It's something being actively worked on, but it's still only half baked and I wouldn't recommend anyone daily drive a Linux phone. Maybe in a few more years it will reach a state where it's actually usable.

One thing that would help a lot is if some company stepped up to provide a platform agnostic NFC payment solution that worked on both iOS and Android. As far as I'm aware if you want NFC payment you have exactly one choice depending on your OS, and both Apple and Google brick NFC if you root your device.

orclev , 9 days ago

Not my entire payment record but certainly everything I use my phone to pay for. I'm willing to give Google some of my info as long as I'm in control of what info I'm giving them. Everything I do on my phone is too much. If a 3rd party offered a NFC payment app I'd happily use that over GPay, but until that exists GPay is the only option. Ultimately GPay is safer than using actual credit cards because it's more resistant to skimming. The extra security outweighs the loss of privacy in this specific case. I'm not happy about that but there doesn't seem to be a better alternative at this time.

orclev , 9 days ago

You know that if someone skims your card and makes a fraudulent purchase, you will likely be able to get your money back, right?

Sure but it's a major pain in the ass. Every time it happens I have to cancel my current cards, request a new one, find all the services I'm currently paying with the now cancelled card and update them to a different card while I wait for the replacement, and then maybe remember to swap them back when the new card shows up. It doesn't happen constantly but if I use cards to pay they seem to get skimmed about once every year or two.

What do you think will happen if someone exploits a 0-day in GPay to do this? How could your bank know the purchase was fraudulent? At least with a card it is obvious that this can happen.

Literally never happened before, but same way they know a credit charge is fraudulent, I tell them. Also if someone found a 0-day in GPay I wouldn't be the only one complaining of fraudulent charges, they'd be flooded with complaints.

If you care about "secure" payments that much, why not use cash?

Because that's a pain in the ass. I don't care about "secure" payments, I care about not having to spend days dealing with the aftermath of it. Paying with cash means I need to constantly go to ATMs to withdraw money, and if I'm doing that my odds of getting my card skimmed actually go up so it doesn't even protect my from that.

orclev , 9 days ago

Technically you're correct, but it's effectively the same thing since I've literally never used NFC for anything besides contactless payment and initial phone setup when migrating from an older Android phone to a newer one. For most people NFC is synonymous with contactless payment.

orclev , 9 days ago

Honestly the short 5 year from original release till EOL thing really fucking annoys me, but it's literally every phone on the market. I've looked, it's impossible to find a phone that doesn't force you to replace it every few years unless you go to a plain dumb phone that only supports voice calls and maybe basic SMS with no apps. That's just a nonstarter in this day and age.

Even alternative Android firmware like GrapheneOS and /e/OS are dependent on the stock firmware releases by the phone manufacturer so when the manufacturer goes EOL and stops releasing updates your alternative installs also are effectively EOL.

The only solution to this problem I've seen that seems like it has a chance is Linux Phone OS, but it still has several problems that make it unusable for most people (biggest one probably being that it provides absolutely terrible battery life).

orclev , 10 days ago

I am ashamed to say I'm an ex-Arch user. I really liked Arch, but if you go long periods without updating it it becomes really painful when you do eventually apply updates. I got tired of finally booting some system I hadn't touched in a year to work on something or other attempting to apply updates to it, and then spending the next 6 hours fixing everything that broke because one of the updates from last week applied before one from 10 months ago that it depended on, or some config file I touched once 3 years ago changed and needs manual fixing now.

orclev , 11 days ago

Never use anything that depends on the Internet for your smart home. There are entirely offline text to speech and voice recognition plugins/libraries for home assistant.

orclev , 11 days ago

It's hilarious you think Apple is in any way pro-consumer. Apple is all about their walled gardens, but they're a trap. Their entire business model is designed to use various underhanded means to entice you into their ecosystem and at every step make it increasingly difficult to escape it all so that they can keep extracting money from you. Google and Microsoft aren't much better but they are better ironically because they're not as good at disguising their bait and traps as Apple is.

orclev , 11 days ago

Depends somewhat. Home Assistant has its own app that works pretty well. Generally other apps don't integrate with Home Assistant, but rather the other way around with Home Assistant having a plug-in to allow it to control and interact with other hardware/software. Everything is controlled/managed by Home Assistant so when you need to issue commands or setup triggers its your one stop shop.

I know there are voice control plugins for Home Assistant that are reviewed reasonably well, but I don't personally use them so I can't say what models of mic work well or not. My usage is pretty minimal with simple timers and/or things like presence and temperature sensors being sufficient to meet all my automation needs supplemented by the occasional custom UI button on my Home Assistant dashboard accessible through the app or web interface. Personally the rule I try to follow for all my IoT devices is to avoid Wifi if at all possible in favor of standard mesh network tech like Zigbee or Z-Wave that way it's literally impossible for my "smart devices" to call out to the Internet without me manually installing some kind of bridging software in Home Assistant.

orclev , 11 days ago (edited 11 days ago)

So I got curious and did a little research. Home assistant itself provides several tutorials for setting up voice controls. There are several options to choose from. The simplest to get up and running seems like signing up for home assistant cloud and paying a monthly fee, although that's also the least private and most costly option. At the other extreme you can go entirely DIY by picking up a cheap dev board for around $14, flashing it with ESPHome and hooking it up to your local voice assistant setup.

I haven't personally used any of this as I've said previously, but everything I've read seems reasonable and it sounds like it's the generally accepted way to go about doing these things with Home Assistant.

Edit: Reading the reviews of the ATOM Echo it seems like a lot of people complain about the mic and speaker quality and issues with it going into sleep mode. A more expensive but possibly better option might be one of the ESP32 S3 Box models flashed with ESPHome, although once again I have not personally used any of this so I can't say whether the complaints about the ATOM Echo hold water or if the S3 Box is better. There also seems to be a Assist app for at least Android, so if you don't want a dedicated hardware device you might be able to make do with just your phone.

Edit 2: This also looks like it might be a great option. It looks a little more polished and ready to go, and is supported by ESPHome.

orclev , 12 days ago

Based on all the ways I've seen AI used recently it means...

checks notes

A computer program did something, possibly while remembering something it did previously.

orclev , 23 days ago

Calling him a religious weirdo gives too much credit to the cult/scam that scientology is. At best he's a brainwashed cult member. I feel like 200 years from now people will be studying the rise and fall of scientology as it's a fascinating case study of what happens when a scammer sets out to create a cult and actually succeeds. The fact he got away with it despite evidence that it was always intended as a scam is even more mind blowing.

orclev , 23 days ago

I mean yeah, but it's interesting that even with all that readily available evidence of how much of a scam it is people still sign up. At the end of the day the only real difference between a cult and a religion is how old it is. But while you can give followers of other religions the benefit of the doubt because that evidence has been lost to time, it's very much still available for scientology. Hence calling him a religious weirdo is too much credit, followers of scientology have ample evidence that it's a scam/cult, but they choose to ignore that evidence. There's basically no excuse for believing in scientology much like there's no excuse for believing the Earth is flat.

orclev , 23 days ago

Honestly the fact it had any CGI was groundbreaking. We take it for granted these days how easy CGI is, but at the time Tron was made movies were still recorded on physical film and most computer monitors were 480P resolution at best. The movies in the 70s and 80s that had "digital" displays like the terrain map in Aliens used some really clever tricks to fake things that would be utterly trivial to do today but were almost impractical to do back then.

orclev , 25 days ago

Yeah, living causes cancer. The real question is, how much does something increase your risk of developing cancer. If it's less than the increase from walking around outside for a few hours on a bright day you can pretty safely ignore that. As long as you're not eating the interior of your car I doubt this poses a significant risk.

orclev , 25 days ago

Synthetic fabrics without flame retardants are basically wearable napalm. You absolutely want flame retardants in your synthetic fabrics because without them if you ever get exposed to a fire you can basically kiss your skin goodbye. If you can't live with the retardants you better start wearing nothing but 100% cotton clothing. I mean you'll still burst into flame pretty much instantly like that, but at least it won't stick to you at the same time.

orclev , 25 days ago

Natural fibers will still combust with shocking speed without flame retardants, they just won't stick to you while doing so. From the moment of being exposed to flame to the point at which the entire garment is on fire can be as little as a few seconds. No matter the increased risk I still rather keep the flame retardants, because death from large burns is a nasty way to go.

orclev , 25 days ago

Well it says people with a high blood concentration of these chemicals have a 4x increase vs. those with a low concentration. That sounds bad but it might not be. If your odds of developing cancer in the low concentration group are 1 in a million, then your odds in the high concentration group are only 1 in 250,000 which isn't exactly great but isn't terrible either. On the other hand if your odds in the low group are 1 in 10,000, then in the high group it's now 1 in 2,500 which is pretty bad.

All that is also ignoring that the article never directly says cars are responsible, only that the chemicals are present in them, and that people with a high blood concentration of those chemicals have a higher risk. Time is also never discussed. Does it take 80 years of near constant exposure to reach "high blood concentrations", or are we talking like 5 years? The article is just too nebulous and vague. It shows some correlations, but seems to fall short of both causal links and quantifying the actual risks.

orclev , 24 days ago

You completely missed the point. 400% is meaningless without more context. If I increase my odds of winning the lottery by 400%, I'd still be a moron for wasting my money in the lottery. Percentages are constantly abused in marketing and news articles to imply things that don't really apply.

So yes, the article doesn't actually specify how much your risk increases due to being exposed to those chemicals, just saying 400% is about as informative as saying 6 or 10,000. It implies a significant risk, but doesn't support it. Without knowing how much risk there actually is it's impossible to evaluate if the benefits outweigh the risks.

orclev , 25 days ago

That's because if they had an actual good reason for doing something they would just say it. If they're using the nebulous "protecting children" they're just trying to invoke a boogeyman and shield themselves from criticism because nobody wants to go on record as being in favor of endangering children which is the implied stance if you oppose "protecting children".

orclev , 1 month ago

EK has been a major player in the water cooling market for a long time. They practically invented full coverage GPU blocks and they regularly place in the top spots for cooling performance of discrete loops (their AIOs aren't bad, but they also have fierce competition in that space).

That said it's disappointing to see how far they've fallen. It seems like EK entrusted running the business to the wrong guy and they're stuck in a death spiral now. It's unfortunate that EK isn't handling the fallout of that better though. If he had come out with a strong response following the firing of the CEO rather than acting like he has nothing to do with it this would all be playing out very differently.

orclev , 1 month ago

If that's all you're doing you could save a $1000+ and just get a cheap Chromebook. Or if you want to be sustainable and reduce e-waste you could spend around the same amount on a framework laptop that's upgradeable and then spend a tiny fraction of that every few years keeping it up to date, rather than going the Apple approach and chucking the whole thing in the trash every few years and buying a brand new one.

No matter how you slice it, an 8GB macbook is a crap deal.

orclev , 1 month ago

What a load of nonsense. You've got no idea how a computer works. RAM isn't just used for passing data between cores. If anything that's more the role of cache although even that isn't strictly accurate.

Whether a system has a discrete GPU or not doesn't really factor into the discussion one way or another, although even if it did having more RAM would be even more important without a discrete GPU because a portion of the system RAM gets utilized as VRAM.

orclev , 1 month ago

In the early days of the internet that was actually the correct way to do things. Websites were supposed to be hosted at the www.thisstupidcompany.comr content was supposed to go to places like ftp.whatever.com. Technically they should redirect thisstupidcompany.com to www.thisstupidcompany.com not the other way around. All that said though that concept has kind of fallen by the wayside as it became increasingly apparent that most hosts exist primarily to host websites with other uses as secondary so it became the norm to just assume the www when navigating to the root domain.

orclev , 1 month ago

Longer term it's going to be interesting to see what if anything RISC-V changes. Right now they're filling a role that ARM occupied about 20 years ago being primarily an alternative for cheap and medium power devices, but just like ARM they've got the potential to duke it out in the desktop space with the right backing. It would for instance be an interesting move if Microsoft partnered with a company like HiFive to produce a truly high end RISC-V CPU similar to Apples M1/M2.

orclev , 1 month ago

Nothing runs on a processor it wasn't built for. That said you can paper over that somewhat using an emulator which is what Apple did on OS X to let you run x86 apps on ARM processors. There's a performance tradeoff though, you obviously can't run as fast in an emulator as you could natively, but if the native processor is fast enough you might not notice the difference. With games in particular they're not often CPU bound, so assuming a good enough GPU even in an emulator it might run just fine. That said games are also far more sensitive to minor timing fluctuations, so even if 90% of the time it runs fine if 10% of the time it has bad hitching you're going to have a bad experience.

For a slightly less extreme example of this you can look at Steamdeck. That's running games in an emulator but in that case it's a more mild form of emulation since it's the same CPU architecture, it's just emulating several APIs to make Linux look like Windows to the game.

orclev , 1 month ago

Yes, I also didn't want to get into the whole "Wine Is Not an Emulator" thing. Technically speaking I suppose it would be most accurate to call it a compatibility shim, although the extremes it goes to somewhat stretch the definition of shim.

orclev , 1 month ago

Yes, technically speaking ARM is RISC, just a different flavor of it from RISC-V. They're effectively siblings. x86 on the other hand (and AMD64) are CISC processors. CISC provides compact programs at the cost of a more complicated (and therefore more power hungry) CPU. That said this is a gross oversimplification and no modern CPU is entirely RISC or CISC under the covers. Both ARM and x86 end up looking quite similar to each other when you dig into them, with x86 producing microcode from its instruction set that is effectively RISC, and ARM introducing some decidedly CISC looking instructions.

The reality is the relative power hungry-ness of the architectures doesn't really come down to RISC vs. CISC as much as it does x86 providing backwards compatibility to literally decades of bad decisions. If x86 could jettison backwards compatibility and ditch all but the latest and greatest of its instruction set it would be able to compete watt for watt with ARM easily, but that's a tradeoff customers are unwilling to engage with as it would render large swaths of software incompatible.

orclev , 1 month ago

Producing a really high end CPU just be muscle flexing. Anybody can do that.
Having apps run on it is a whole another story.

You say that, but nobody has actually done so. HiFive has produced some CPUs that would qualify as extremely low end desktop CPUs, but nothing that can compete with even middle of the road processors like an i5 or a Ryzen 5. As for apps, it would be pretty trivial to get a huge swath of Linux apps running on it, and if there was enough of a base and demand you'd see companies producing RISC-V binaries as well (much like they're starting to for ARM). For emulation layers I'm sure something could be done, QEMU if nothing else could probably be used.

orclev , 1 month ago

Hey, I had to change the toner in mine!... once... after like a decade.

orclev , 1 month ago

My bigger question is WTF is this system even doing? If you win money in the lottery, you can select to apply it to a vacation package if your random draw hits it?

No, it's advertising. They're trying to convince people to play the lottery so they have you roll a (virtual) wheel and upload a head shot then it generates a theoretical video of what it might look like if you went on that vacation (using your theoretical future winnings). It's absolutely idiotic, but their target demographic isn't exactly the sharpest tools in the shed to begin with.

orclev , 2 months ago

Right but you're missing the point. It doesn't matter if 1 BTC costs $1,000,000 when you can just buy 0.0001 BTC for $100. They would need to drive the cost of 1 BTC into the quadrillions of dollars before it would become even remotely unfeasible to purchase.

orclev , 2 months ago

It's the CIA, that kind of thing would probably get you put in the priority recruitment pool. That's practically their bread and butter.

orclev , 2 months ago

Hmm, if it doesn't honor that setting being changed after the initial install it could be possible to set it during install to get the benefits, then change it post install to make other apps behave normally.

orclev , 2 months ago

Yeah pretty much this. My understanding of the way LLMs function is that they operate on statistical associations of words which would amount to categories in Category Theory. Basically the training phase is classifying words into categories based on the examples in the training input. Then when you feed it a prompt it just uses those categories to parse and "solve" your prompt. It's not "mysterious" it's just opaque because it's an incredibly complicated model. Exactly the sort of thing that people are really bad at working with, but which computers are really good with.

orclev , 2 months ago

Most likely the terms say that you agree to go through individual binding arbitration rather than a lawsuit which the courts have found to be legal and enforceable. It's really shitty and has become corporations favorite weapon to use against people, particularly because the arbitration companies are usually fairly friendly towards whatever corporation is being challenged. Contractually mandated arbitration really needs to be invalidated. Arbitration is a fine alternative if both parties want to go that route but it should never be forced on someone, particularly because of some bullshit EULA.

orclev , 3 months ago

Yep. DRM has been and continues to be a complete waste of everyone's time that only makes things worse for paying customers. Pirates get the best experience and then companies wonder why they struggle to get people to pay for inferior experiences. Gabe Newell hit the nail on the head over a decade ago when he said:

The easiest way to stop piracy is not by putting antipiracy technology to work. It's by giving those people a service that's better than what they're receiving from the pirates.

Instead companies keep doing the exact opposite, and surprise piracy isn't impacted at all.

orclev , 3 months ago

It is reductive to say that piracy is just a service problem. There are lots of people who will try to save the money. But a lot of those people wouldn't spend much if any money either way. They would just skip most content, or watch with friends or similar. There is a huge group of people (myself included) that would happily pay a significant amount for content if they provided a good experience. But they are too busy failing to stop piracy to bother giving a good experience.

Yeah I mean you've basically got three district groups at play.

The first group, either have no money or no interest in your goods or services. They might turn to piracy if it's available, but even if it isn't they're still not buying anything from you. DRM is pointless to this group because it's not stopping anything.

The second group are the marginal cases. They potentially have the money to buy your products, but maybe they're pinching pennies or they aren't convinced your products are worth the price you're asking for them. A lot of pirates of Adobe PhotoShop a couple decades back would have fallen into this group. DRM might be effective on this group, but there's a strong argument to be made that it's going to cost you just as many sales as it earns you, and ultimately doesn't actually stop piracy, merely delays it a bit. You'd likely see as many or more sales from this group if you removed the DRM and added more features or cut your prices

The last group are your paying customers. They're already happily (or at least grudgingly) giving you money. The only thing DRM is doing for this group is making their experience worse and likely pushing them towards that second group.

There's basically no group where DRM is really improving things. At best you're breaking even, at worst it's costing you sales, to say nothing of the development costs of implementing the DRM in the first place.

orclev , 3 months ago

For me it's the "l".

orclev , 3 months ago

It seems like Arc is popular among users of Chrome, but there's little reason for anyone already on a better browser to use it. Every single person I've heard saying good things about Arc is a long time Chrome user.

orclev , 3 months ago

Because you very clearly don't understand how LLMs work and are describing something that's impossible. If you did have something that worked like that it wouldn't be a LLM, it would be something fundamentally different and closer to a true AI. There are no true AI in existence currently, and just trying to train a LLM using its inputs won't change that, it would just make the output worse by introducing noise.

orclev , 3 months ago

That would embolden the anti-trans crowd.

Unfortunately no matter the outcome there's likely danger there which is why legal protections are going to be critical. If it doesn't accurately detect trans people they'd argue it's evidence that being trans is a choice and that they could just decide to "be normal". If it does accurately detect trans people in some ways that's even more dangerous because now you've created a trans detector that could potentially be used to target people.

Ultimately from a scientific perspective there's still so much we don't know about how our brains work and even less known about how gender and sexual orientation are determined. Projects like this provide valuable clues about all of that, but there's still so much that's unknown that any result is potentially useful. I personally would find it more interesting though if it did accurately detect someone whose trans as it would suggest there's physically detectable brain differences.