Exactly. If someone is exploiting this in my house, then it means they already broke in and have complete physical access to my house, screwing around with the Echo and maybe making fradulent Amazon orders or whatever would be the least of my concerns.
Yeah, I'm struggling to see the use case. Maybe a cloak-and-dagger situation where you have limited legitimate access under close scrutiny and want to plant a bug but can't do it physically, like say you're a fake inspector at a drug lord's house. All you have to do is make some pretext to walk past the device with the ultrasonic command playing and it'll go to some malware site and root itself. Pretty far fetched though...
On the other hand, it would only take a single bug in an AI "debris subroutine" running in a whole bunch of self-driving cars to choose the wrong course of action 100% of the time. Such a bug would *probably* only be identified after enough failures were accurately recorded to piece together a pattern that could point to it
What makes you say that? I would expect any SDC accident or near-accident where the car is potentially at fault to be given a thorough hearing, FAA style. I expect the "black box" in SDCs to give you all the raw sensor data from the last 30 seconds which will be put into simulators and ran not just on what did happen but a ton of variations to see what could happen. And for any change in the programming to be ran through a bunch of regression tests to check that you don't have any unexpected behavior changes or side effects.
Humans make individual errors and there's only so much point in studying them, like you say SDCs should have predictable behavior in that they all work the same and should react in exactly the same way to exactly the same input. So you pounce on the first instance that it might have a problem and fix it, also updating all other cars on the road who haven't ran into that oddball combination of circumstances that caused a problem before there's ever a second accident. Not that you'll ever have 100% coverage or that there won't be bad patches, but I see this as inherently much better.
Every year there's a new age cohort of drivers on the road with zero experience and they make newbie mistakes, so was I once. And even with experience we're not 100% concentrated on driving like a race driver, we have small lapses because we're distracted, tired, emotionally unbalanced and so on. I don't expect perfection from a SDC, but computers tend to have exceptional consistency. Of course there's too many variables for every day to be exactly the same, but at least the car will drive the same way, every time.
If you break it down to "micro-situations" maybe there aren't that many different algorithms, because you only have to identify the situation sufficiently to reach a conclusion. If it's a drunk man or person in wheelchair or child or pet or wild animal in the road all amounts to animate object in road, stop. Also the rules for passing through an intersection are pretty universal, you have to identify the specifics of traffic lights and cars and pedestrians and whatnot, but I expect that to be one "thing" that's tweaked and refined to work in 99.9% of intersections 99.9% of the time. And the reminder it'll just be a stopped idiot.
I doubt that this can ever be a criminal trial, because no criminal intent is involved at any level of the design or implementation of the self driving car. It's an accident.
The two highest levels of culpability, purposely or knowingly probably not. The two lowest levels, recklessness and negligence can most certainly happen. The former would be where the court finds that there was a programming decision made to ignore a potentially dangerous condition, the latter if it failed to recognize the condition or react in a reasonable manner. It's unlikely that any person will be held criminally liable, unless it's a matter of deliberate cheating like the emissions scandal but the company can certainly see a criminal prosecution. For them it's not that different from a civil case though, it's all just money. Though if the court revokes the company's driving license, it could be a lot of money...
IMO, Freenet was the right approach - turn everything into a big distributed hash table. (...) Too bad it's so inherently slow it never took off
I don't think it was inherent, it was just stupid. I looked into it... wow, it must be 10 years ago or something and it didn't scale because with many nodes and relatively small node sizes to total data it'd almost be an accident if you found what you looked for. Increasing your own node's connections helped massively, which it shouldn't. I was studying the effect of each node not only picking a place in the hash table, but also creating a highly directed network with more slots the closer your hashes. That way you'd be much more likely to get closer and closer instead of running into dead ends or circle loops as Freenet tended to do. That created a very strong "gravity" to the network that could be analyzed but I essentially divided the problem into two parts, in the beginning the request bounced randomly for anonymity then the "homing" kicked in and found the content for efficiency.
As part of that I also created the opportunity for "shortcuts", like if it had gone node 1-2-3-4-5-6-7-8-9-10-11-12-13 to find the content node 3 could add a "and here's my contact info" and node 12 maybe pick up on it so the return path was 13-12-3-2-1, either temporarily for that one block or as a new link in the network. That created the opportunity for much faster downloads, in particular if you as a downloader didn't care you could shout "hey, here I am" to everyone for a torrent-like download. That of course wouldn't work for everyone but the risk is often asymmetrical or depends on your location/situation while Freenet has only one speed for everything in both directions.
The main problem though was that it didn't do onion routing which made it extremely vulnerable to statistical analysis and because of its terrible algorithm for locating anything routing through a random node was like finding a needle in a haystack. With my algorithm that started to seem plausible though, you could actually send a request to a particular node without knowing the path, that would retrieve the content and forward it to you encrypted. Or at least it could in a simulation of a million nodes with 50 connections/node. It also made onion inserts possible, which is probably even more important. Also data was padded to half-powers of 2, so if you were sending/receiving an image 166321 bytes long it would be 192kb like everything else from 128kb to 192k. Plus I experimented with various kind of mix-master buffers and fake traffic to make traffic analysis difficult to say where any stream actually begins, ends or is routed to.
With anonymous inserts you could also do tricks like double keys (XOR the content, hash again) that would render "denial-of-keyspace" attacks useless since one of the two keys would be good. If you trust the onion routing you could do automatic healing with PAR blocks meaning files are much more likely to download fast or not at all, with Freenet that was a pretty big security risk. Also I experimented with insert-specific keys for the same data so you could insert known files with lower risk. Also a more intelligent system for signed pointers so you could do RSS-like feeds. I dabbled in making a forum/email system but never found a true cure to spam, enough trash could bring anything down. I also managed to make peer circles, that was a pretty neat anti-analysis trick for low-volume content... basically the same block goes in a circle, only the client and the meeting point alters data, could do encrypted chat with reasonably low lag, not instant but like sub-second.
The one thing I never really got solved was the problem of poison nodes and "catch-and-release"... basically if your node keeps getting offers from one poison node to connect to more poison nodes eventually you'll end up talking only to poison nodes, even with onion routing those too will be compromised. I tried making something equivalent of guard nodes and that nodes you
I view immigration as a case of if your country sucks fix it. If you can't fix it then don't take your messed up ways with you and assimilate into wherever you go.
Sounds to me like you've got nobody but yourself to care about, they're the only ones who think it's just to pick up a gun and start shooting back. The vast majority of civilians are way too busy trying to protect their family and those closest to them. Particularly when it's so unclear who's the good guys and the bad guys and the front lines are so unclear, it's one thing to go to war and leave your family behind in relative safety. It's quite different when they're in the war zone or even occupied territory with you, against an enemy that obey no rules of war. Millions of people have fled for good reasons to the surrounding areas and I don't blame them one bit for doing that.
What doesn't add up is that almost all the migrants that come to Europe are young and male. Basically you're not sending women and children, you're sending the young, strong and "expendable" males of the family ideally 15-20yo so they can claim to be underage after burning all their papers. And they are supposed to be the economic or geographic anchor to get the rest of the family money or residency. If it looks like they're there to suck the government's teat as hard as they can, that's exactly the mission they were sent for. And they were probably the ones who least needed to leave, it's bullshit. You also don't need to be a genius to realize that many of these will become problem teens/young adults.
Basically, we're helping the wrong people. And we're rewarding smugglers with boatloads (pun intended) of money to do it. And I don't mean it just as a NIMBY defense, save children <10yo and families particularly mothers. They're not the ones causing trouble.
Automation of the service industry sends these workers...nowhere.
The service industry is a pretty big place, an automatic taxi driver yes... an automatic nurse, nah. Not without 50 years of robotics and "I, robot" levels of human interaction. I also expect we'll need some kind of domain experts to work with the robots, like hiring an electrician to do the job and he tells the robot how to do the wiring. It's not like anyone has been able to make business users design sensible IT systems, why would the real world be any different? There's only so much computers can do to stop people from shooting themselves in the foot until they have strong AI and start figuring out what the customer really wants.
No, I grok it just fine and I think we're perfectly in agreement on how this works.
"Downloads are prohibited without prior permission from the Technology Administrator."
Not just applications, but downloads in general? Am I in violation if I download a PDF?
"Employees will not use personal technology at work (...)
So if I check my personal cell phone while at work...
and will not make changes to any of the Firm's technology without prior permission from the Technology Administrator."
I can't even parse this, am I allowed to turn on/off my computer?
She was on match.com (this was the trigger for the firewall block, per my recommendation) on a Friday from 2 pm to 5 pm.
And you religiously enforce this for everyone who spent two minutes checking a non-work related item?
It was all documented, signed by her, and she was let go.
Which was my point.. it's not a policy you expect people to follow, it's a policy everyone violates so you can fire those you want to fire who have violated your *real* thresholds for unacceptable behavior.
I think you're missing the key issue here which is that even without labor things cost money, that Big Mac isn't conjured out of thin air. With higher automation more of money flows to the corporate owners and less back to the labor force through wages. The purchasing power gained by paying less isn't enough to offset the purchasing power lost to unemployment. It's essentially the same argument some make about cheap goods from Amazon/Wal-Mart/China, it kills the local economy. Except in this case we're paying the machines' owners and killing the labor economy. Don't forget that the best productivity possible is a divide by zero...
I got a few calls regarding wrongful termination during the years and, in one matter, the fired employee said, "Well, everyone else was doing it."
I told the work comp lady to add, "Line item 6.1.a, 'Report any violations or suspected violations of this policy to the Technology Administrator."
So assuming he wasn't exaggerating you amended a policy nobody followed with another over-the-top rule for them to ignore, brilliant. I've read a few policies like that, in theory they're great. In practice nobody knows, because they're so anal the only real purpose they serve is as legal ammunition against troublesome employees. For example I read my organization's phone application guidelines, install any non-IT approved app and you take full legal liability for any damage it can cause. Meanwhile using it as your personal phone too is encouraged and 95%+ do exactly that, nobody bats an eye at installing anything. It's only there because if shit hits the fan they can throw you to the wolves and blame you for violating policy.
On a modern machine, you don't need carefully optimised queries and carefully designed on-disk data structures - you can probably fit the whole thing in RAM and run the computations in Lua and still get the whole thing done in a second or two. At one end of the market, most customers can now run their systems with cheap commodity hardware and software. At the other end of the market, companies like Facebook and Google have more data than Oracle can easily handle and couldn't afford Oracle license fees even if there was a viable Oracle product for them to buy. The middle is gradually shrinking.
Well for companies like Facebook and Google you don't care that the social feed or search index is only kinda updated by computer standards or that different people see slightly alternate versions. There are a lot of markets where you don't have that luxury, that ticket is either sold or not sold. But yes I agree those kinds of transaction processing markets don't scale the way computer hardware does, my local cinema got roughly as many seats as it had last decade. Stadiums and arenas don't get bigger. Even financial transactions (outside high frequency trading) and Amazon's inventory don't increase that exponentially. Most of the real crunching power is now going into "softer" processing of big data, voice recognition, robot vision etc.
I do freelance/contract/consulting work. However, I never work through an agency. I always form an agreement directly with my client. In my case I am responsible for all my own taxes along with everything else, which is how I prefer it to be. Is this approach not common? Is it more common for IT/tech/software dev consultants/freelancers/contractors to use an agency? If so, what is the benefit?
You don't need a network, you don't need to be a salesman, you don't have to know all the paperwork and you got a guaranteed base salary. At least that was roughly my experience starting out as a consultant, "all" I had to do was show up at clients and deliver well. As an consulting company they come across many opportunities big and small with different requirements, they were quite good at keeping billable hours high at good rates. It was employee-ish but with more variable pay, more varying projects, technologies and people. If you had downtime you could take classes and certifications. Overall a pretty good fast track to build a good CV.
If you mean agencies that take a cut just to match up the work with the people then I think people try to avoid them. Before you needed them to find the resources but with online database it's much easier to find what you're looking for. There's still those who do all that "other" work for a cut though, basically it's more like you buy a support service but if you don't find your own billable work you're not making any money. Often it's worth it to spend time billing and not dealing with everything else. You can waste a lot of time fiddling with paperwork and when you start to put a price on those hours it might make sense. Unless you're the jack-of-all-trades kind that can both sell and deliver and support yourself and be good at all three.
This is the perfect opportunity to move to torrents over darknets such as I2P (but not Tor). This is also the end-state of piracy, unstoppable and untraceable file sharing. It's the last platform switch you'd have to make. The only downside is that it would force media companies to begin an assault on general-purpose computing.
Not worth the effort. The only place they really bother to try holding back the tide are the torrent websites, once you have the magnet link it's such a giant whack-a-mole. The mass lawsuits have gone nowhere, the "six strikes" policy is just fluff, if you want to be the original seeder maybe you need a VPN but... here's 90 million moles to whack. And that's presumably grossly under-reporting the number of private ways people have shared the file. Even if they managed to prosecute and fine 90,000 of those - unlikely - there's like 0.1% chance of that being you. Even if you get hit with the US statutory minimum of $750 for willful infringement - also unlikely - that's like $0.75 risk-adjusted.
In most other countries you can only sue for actual damages + costs and most places an IP address is also not good enough to win a case in a multi-person household, guests, tenants etc. it doesn't stick to the subscriber. So far the advice here in Norway to these "blackmail letters" is to simply ignore them. Don't acknowledge them, don't admit to anything, don't pay anything and they will go away. They just hope that at least some people will basically incriminate themselves or pay to make it go away. That said, a lot of people do pay when there's decent online services.
You're thinking of the incremental advances from current AI. That will certainly be leveraged, but eventually we will come up with general AI in a way which can be accomplished using available resources. That's decades away according to most people, but any country that develops it first can literally out-think the others in everything, unless they don't have enough lead time. Every government in the world would go to war for that power or to keep that power out of the hands of another.
Replace AI with nukes and out-think with out-kill and the rest of the world should have allied and invaded the US in 1945. You're also assuming that a super-intelligence will appear out of nowhere and that country won't build up to a golden age of economic and industrial power on the way. And that said nation won't ally itself with partner states that'll stand in its halo rather than join a conspiracy to dethrone them. For that matter, the assumption that it'll be a nation state is dubious and not some mega-corporation looking for a generic business optimization engine. Oh yeah, you're also assuming a general engine will be better than a specialized engine. The human brain is amazingly flexible, but it's decades since we lost to computers in chess.
The odd part about that ruling is that it established that APIs are copyrighted, but didn't come up with a single plausible theory of how they could possibly be used in a way that is infringing. If you can copy APIs wholesale to create a competing commercial product you're pretty much in the worst corner of on at least two if not three corners of the fair use test. If the functional nature of interfaces demand that they have to look the same to work the same and will grant you a fair use defense every time it's just semantics.
I'll wait until they've done a much larger properly blinded test.
If I had terminal cancer I'd take highly experimental over certain death, you're only risk averse when it usually ends well. When it'll end badly you're ready for any "Hail Mary" save. Not that I'd try obvious snake oil and superstition, but any reasonable experiment I'd be in on... seems like the worst that can happen here is that you get mighty drunk, granted I've had bad hangovers but I'd rather go out drunk as a skunk than wait for the cancer to get me.
The only way around that is to use some technique that trims the blockchain, but I'm not sure if that's even technically possible.
Sure it would, just have the blockchain sign a complete account balance and everyone could restart from there. It could be vulnerable to a 51% attack totally rewriting the balance but you probably also make some sort of rejection mechanism saying that if >1% of the blockchain rejects the account summary you continue from the last universally acknowledged one. And some nodes would probably stay in full history mode anyway. That doesn't seem like a particularly hard problem, compared to all the other ones...
Yeah, but you can eliminate that from the equation because 20/20 vision = one arch-minute (1/60th of a degree) of resolution. So if the TV covers 30 degrees field of vision (FOV), you can see 30*60 = 1800 pixels of resolution (with a perfectly curved screen where you don't have aliasing effects, 3600 if you do). Now most young people have better than 20/20 vision because that's just the threshold we've set for normal, doesn't need correction vision and if you're really enjoying your widescreen you might be pushing 50 degrees though that's uncomfortably close for most. But even if you have a 50 degree FOV and 20/16 vision you'll at most see 50*60*20/16 = 3750 pixels, which is like front row at the cinema with your average teen eyesight.
Exceptional people have all the way down to 20/10 eyesight, though 20/8 is the physical limit. Exceptional people might recognize aliasing, in theory you could arrive at a number where you need 15K resolution (50*60*20/8*2) to display a "perfect" image. If you are a photographer and want to like stare at a detail on the screen ignoring the big picture, a 32K monitor might make sense. But big Hollywood productions like "The Martian" have made a 2.5K master and nobody has complained, truth is even 4K seems overkill for video. It's not like anybody can actually hear the difference between 24 bit / 96 kHz and 16 bit / 44.1 kHz, it's all in their heads. Good for editing but makes zero difference for presentation.
All you are doing is extending the value to other things that also don't have value and saying that is why it has value. Everything in the crypto-currency markets have the same problem. Just because one pump and dump stock is used to trace between other pump and dump stocks doesn't make any one of them valuable.
That's what *all* currencies are about... are there going to be telltale signs of people retreating to "safe" currencies... there's a reason why in times of trouble people invest in USD, swiss francs (I forget the TLA) and EUR it's because they're massive economies that won't fail without an economic Armageddon. Also precious metals like gold and silver that people assume will "always" be valuable without any real reason except for thousands of years of history.
While the price of bitcoin keeps surging people won't use it for things outside of the crypto-currency markets. It must stabilize, and the entire design of it won't allow that. It will either keep surging or it will fall, stable is not what it is designed for.
You're casually dismissing all the actual things bought and sold with bitcoin like they don't exist? Even if it happens to be illegal substances or various other dubious things they're real, as in people who don't give a shit about speculation but who buy bitcoins to buy it. And the dealers sell for "real money" too, as they can sell them to new buyers. They don't care what the going rate is, it's just a proxy. And between speculators and users you have a working currency, whether you deny it or not.
V2E communication is the absolutely least thing needed for autonomous vehicles. Identifying other vehicles, traffic lights and road signs is completely trivial compared to dealing with "everything else". Random people, animals, debris and whatnot will never wear a chip, if one car can't deal with it on its own neither can a group. The potential benefits of "near-psychic" autonomous cars moving as a pseudo-train and timing the crossing of the intersection is nice-to-have to optimize cars that are already self-driving.
A 30% jump in one month, after two decades of "YYYY will be the year of Linux on the desktop!" ? The explanation is obvious: bad data.
Almost certainly, it would take an extreme swing in sales/installations to make the number of active Linux devices swing 30% in a month. After all, the vast majority of people use the same OS this month as they used last month. If you say people make an OS choice every 3 years, you'd at most expect 100%/(12 months/years*3 years) = 2.8% change per month even if everybody switched to a new OS. Also it's not collaborated by other sources, the desktop market share on statcounter has barely moved, Linux has gone from 1.74% to 1.79%. When these curves swing it's because there's been some kind of change in how they identify devices and platforms, either because they changed the definition or because new devices challenge it.
... store everything I've ever written or photographed on one of these. Every personal record, bank statement, tax form. Then sneeze once and its lost in the shag rug forever.
Just ask the NSA for their copy... for bonus point they'll also have all your phone calls transcribed, your photos will be marked with facial recognition and location recognition and cross indexed with the cell phone tower, GPS and open WiFi name records of where you've been and every http URL you've ever visited. And I'm only half joking, it's creepy that technology now actually makes it feasible to store practically everything about practically everyone.
I've always been an AMD fan, but their offerings have been lackluster lately. On my gaming rig I finally went Intel and on my regular desktop machine I'm still running AMD but it's an ancient 7-8 year old Phenom II. I haven't had much reason to upgrade it until now but I'm thinking I may finally pull the trigger on a system built on the "budget" Threadripper.
Note that the budget Threadripper is a 4+4 system with non-uniform memory, if you want it for the PCIe lanes (64), memory bandwidth (quad channel) or memory size (8x16GB = 128GB possible) fine, but it'll have higher memory latency and the idle power consumption is much higher because you have 2 dies + an interconnect that's always on. It'll probably still do slightly better than an 1800X because it clocks higher since the TR platform has higher TDP and beefier cooling but the motherboards are also likely to be more expensive because of all the extra traces.
I think this product is rather niche, most people would be better off with a Ryzen and those who need the TR features often want more cores too. It's a pretty good deal for AMD, they can take two "junk" Ryzen processors with 50% defective cores and sell at higher price/core than a full 1800X with 8 of 8 cores working. If you need the unique features of TR it's still a good deal, but I don't think it's a processor you should casually buy without a rather specific plan as to how you're going to use it.
You forgot the Nissan Leaf, it's been the #1 seller for electric car here in Norway for some time. Though it's definitively been a #2 car for most families, it's not exactly made for road trips. Or any long trip really, it's a commuter car to do you daily chores around town. It works for some as long as you got a family car for the other trips.
Slashdot Mirror
Can't you use the A/B support for that, one for stock Android, one for custom OS? From the example they seem to have different rollback indexes.
Exactly. If someone is exploiting this in my house, then it means they already broke in and have complete physical access to my house, screwing around with the Echo and maybe making fradulent Amazon orders or whatever would be the least of my concerns.
Yeah, I'm struggling to see the use case. Maybe a cloak-and-dagger situation where you have limited legitimate access under close scrutiny and want to plant a bug but can't do it physically, like say you're a fake inspector at a drug lord's house. All you have to do is make some pretext to walk past the device with the ultrasonic command playing and it'll go to some malware site and root itself. Pretty far fetched though...
On the other hand, it would only take a single bug in an AI "debris subroutine" running in a whole bunch of self-driving cars to choose the wrong course of action 100% of the time. Such a bug would *probably* only be identified after enough failures were accurately recorded to piece together a pattern that could point to it
What makes you say that? I would expect any SDC accident or near-accident where the car is potentially at fault to be given a thorough hearing, FAA style. I expect the "black box" in SDCs to give you all the raw sensor data from the last 30 seconds which will be put into simulators and ran not just on what did happen but a ton of variations to see what could happen. And for any change in the programming to be ran through a bunch of regression tests to check that you don't have any unexpected behavior changes or side effects.
Humans make individual errors and there's only so much point in studying them, like you say SDCs should have predictable behavior in that they all work the same and should react in exactly the same way to exactly the same input. So you pounce on the first instance that it might have a problem and fix it, also updating all other cars on the road who haven't ran into that oddball combination of circumstances that caused a problem before there's ever a second accident. Not that you'll ever have 100% coverage or that there won't be bad patches, but I see this as inherently much better.
Every year there's a new age cohort of drivers on the road with zero experience and they make newbie mistakes, so was I once. And even with experience we're not 100% concentrated on driving like a race driver, we have small lapses because we're distracted, tired, emotionally unbalanced and so on. I don't expect perfection from a SDC, but computers tend to have exceptional consistency. Of course there's too many variables for every day to be exactly the same, but at least the car will drive the same way, every time.
If you break it down to "micro-situations" maybe there aren't that many different algorithms, because you only have to identify the situation sufficiently to reach a conclusion. If it's a drunk man or person in wheelchair or child or pet or wild animal in the road all amounts to animate object in road, stop. Also the rules for passing through an intersection are pretty universal, you have to identify the specifics of traffic lights and cars and pedestrians and whatnot, but I expect that to be one "thing" that's tweaked and refined to work in 99.9% of intersections 99.9% of the time. And the reminder it'll just be a stopped idiot.
I doubt that this can ever be a criminal trial, because no criminal intent is involved at any level of the design or implementation of the self driving car. It's an accident.
The two highest levels of culpability, purposely or knowingly probably not. The two lowest levels, recklessness and negligence can most certainly happen. The former would be where the court finds that there was a programming decision made to ignore a potentially dangerous condition, the latter if it failed to recognize the condition or react in a reasonable manner. It's unlikely that any person will be held criminally liable, unless it's a matter of deliberate cheating like the emissions scandal but the company can certainly see a criminal prosecution. For them it's not that different from a civil case though, it's all just money. Though if the court revokes the company's driving license, it could be a lot of money...
IMO, Freenet was the right approach - turn everything into a big distributed hash table. (...) Too bad it's so inherently slow it never took off
I don't think it was inherent, it was just stupid. I looked into it... wow, it must be 10 years ago or something and it didn't scale because with many nodes and relatively small node sizes to total data it'd almost be an accident if you found what you looked for. Increasing your own node's connections helped massively, which it shouldn't. I was studying the effect of each node not only picking a place in the hash table, but also creating a highly directed network with more slots the closer your hashes. That way you'd be much more likely to get closer and closer instead of running into dead ends or circle loops as Freenet tended to do. That created a very strong "gravity" to the network that could be analyzed but I essentially divided the problem into two parts, in the beginning the request bounced randomly for anonymity then the "homing" kicked in and found the content for efficiency.
As part of that I also created the opportunity for "shortcuts", like if it had gone node 1-2-3-4-5-6-7-8-9-10-11-12-13 to find the content node 3 could add a "and here's my contact info" and node 12 maybe pick up on it so the return path was 13-12-3-2-1, either temporarily for that one block or as a new link in the network. That created the opportunity for much faster downloads, in particular if you as a downloader didn't care you could shout "hey, here I am" to everyone for a torrent-like download. That of course wouldn't work for everyone but the risk is often asymmetrical or depends on your location/situation while Freenet has only one speed for everything in both directions.
The main problem though was that it didn't do onion routing which made it extremely vulnerable to statistical analysis and because of its terrible algorithm for locating anything routing through a random node was like finding a needle in a haystack. With my algorithm that started to seem plausible though, you could actually send a request to a particular node without knowing the path, that would retrieve the content and forward it to you encrypted. Or at least it could in a simulation of a million nodes with 50 connections/node. It also made onion inserts possible, which is probably even more important. Also data was padded to half-powers of 2, so if you were sending/receiving an image 166321 bytes long it would be 192kb like everything else from 128kb to 192k. Plus I experimented with various kind of mix-master buffers and fake traffic to make traffic analysis difficult to say where any stream actually begins, ends or is routed to.
With anonymous inserts you could also do tricks like double keys (XOR the content, hash again) that would render "denial-of-keyspace" attacks useless since one of the two keys would be good. If you trust the onion routing you could do automatic healing with PAR blocks meaning files are much more likely to download fast or not at all, with Freenet that was a pretty big security risk. Also I experimented with insert-specific keys for the same data so you could insert known files with lower risk. Also a more intelligent system for signed pointers so you could do RSS-like feeds. I dabbled in making a forum/email system but never found a true cure to spam, enough trash could bring anything down. I also managed to make peer circles, that was a pretty neat anti-analysis trick for low-volume content... basically the same block goes in a circle, only the client and the meeting point alters data, could do encrypted chat with reasonably low lag, not instant but like sub-second.
The one thing I never really got solved was the problem of poison nodes and "catch-and-release"... basically if your node keeps getting offers from one poison node to connect to more poison nodes eventually you'll end up talking only to poison nodes, even with onion routing those too will be compromised. I tried making something equivalent of guard nodes and that nodes you
I view immigration as a case of if your country sucks fix it. If you can't fix it then don't take your messed up ways with you and assimilate into wherever you go.
Sounds to me like you've got nobody but yourself to care about, they're the only ones who think it's just to pick up a gun and start shooting back. The vast majority of civilians are way too busy trying to protect their family and those closest to them. Particularly when it's so unclear who's the good guys and the bad guys and the front lines are so unclear, it's one thing to go to war and leave your family behind in relative safety. It's quite different when they're in the war zone or even occupied territory with you, against an enemy that obey no rules of war. Millions of people have fled for good reasons to the surrounding areas and I don't blame them one bit for doing that.
What doesn't add up is that almost all the migrants that come to Europe are young and male. Basically you're not sending women and children, you're sending the young, strong and "expendable" males of the family ideally 15-20yo so they can claim to be underage after burning all their papers. And they are supposed to be the economic or geographic anchor to get the rest of the family money or residency. If it looks like they're there to suck the government's teat as hard as they can, that's exactly the mission they were sent for. And they were probably the ones who least needed to leave, it's bullshit. You also don't need to be a genius to realize that many of these will become problem teens/young adults.
Basically, we're helping the wrong people. And we're rewarding smugglers with boatloads (pun intended) of money to do it. And I don't mean it just as a NIMBY defense, save children <10yo and families particularly mothers. They're not the ones causing trouble.
Automation of the service industry sends these workers...nowhere.
The service industry is a pretty big place, an automatic taxi driver yes... an automatic nurse, nah. Not without 50 years of robotics and "I, robot" levels of human interaction. I also expect we'll need some kind of domain experts to work with the robots, like hiring an electrician to do the job and he tells the robot how to do the wiring. It's not like anyone has been able to make business users design sensible IT systems, why would the real world be any different? There's only so much computers can do to stop people from shooting themselves in the foot until they have strong AI and start figuring out what the customer really wants.
No, I grok it just fine and I think we're perfectly in agreement on how this works.
"Downloads are prohibited without prior permission from the Technology Administrator."
Not just applications, but downloads in general? Am I in violation if I download a PDF?
"Employees will not use personal technology at work (...)
So if I check my personal cell phone while at work...
and will not make changes to any of the Firm's technology without prior permission from the Technology Administrator."
I can't even parse this, am I allowed to turn on/off my computer?
She was on match.com (this was the trigger for the firewall block, per my recommendation) on a Friday from 2 pm to 5 pm.
And you religiously enforce this for everyone who spent two minutes checking a non-work related item?
It was all documented, signed by her, and she was let go.
Which was my point.. it's not a policy you expect people to follow, it's a policy everyone violates so you can fire those you want to fire who have violated your *real* thresholds for unacceptable behavior.
I think you're missing the key issue here which is that even without labor things cost money, that Big Mac isn't conjured out of thin air. With higher automation more of money flows to the corporate owners and less back to the labor force through wages. The purchasing power gained by paying less isn't enough to offset the purchasing power lost to unemployment. It's essentially the same argument some make about cheap goods from Amazon/Wal-Mart/China, it kills the local economy. Except in this case we're paying the machines' owners and killing the labor economy. Don't forget that the best productivity possible is a divide by zero...
I got a few calls regarding wrongful termination during the years and, in one matter, the fired employee said, "Well, everyone else was doing it."
I told the work comp lady to add, "Line item 6.1.a, 'Report any violations or suspected violations of this policy to the Technology Administrator."
So assuming he wasn't exaggerating you amended a policy nobody followed with another over-the-top rule for them to ignore, brilliant. I've read a few policies like that, in theory they're great. In practice nobody knows, because they're so anal the only real purpose they serve is as legal ammunition against troublesome employees. For example I read my organization's phone application guidelines, install any non-IT approved app and you take full legal liability for any damage it can cause. Meanwhile using it as your personal phone too is encouraged and 95%+ do exactly that, nobody bats an eye at installing anything. It's only there because if shit hits the fan they can throw you to the wolves and blame you for violating policy.
On a modern machine, you don't need carefully optimised queries and carefully designed on-disk data structures - you can probably fit the whole thing in RAM and run the computations in Lua and still get the whole thing done in a second or two. At one end of the market, most customers can now run their systems with cheap commodity hardware and software. At the other end of the market, companies like Facebook and Google have more data than Oracle can easily handle and couldn't afford Oracle license fees even if there was a viable Oracle product for them to buy. The middle is gradually shrinking.
Well for companies like Facebook and Google you don't care that the social feed or search index is only kinda updated by computer standards or that different people see slightly alternate versions. There are a lot of markets where you don't have that luxury, that ticket is either sold or not sold. But yes I agree those kinds of transaction processing markets don't scale the way computer hardware does, my local cinema got roughly as many seats as it had last decade. Stadiums and arenas don't get bigger. Even financial transactions (outside high frequency trading) and Amazon's inventory don't increase that exponentially. Most of the real crunching power is now going into "softer" processing of big data, voice recognition, robot vision etc.
I do freelance/contract/consulting work. However, I never work through an agency. I always form an agreement directly with my client. In my case I am responsible for all my own taxes along with everything else, which is how I prefer it to be. Is this approach not common? Is it more common for IT/tech/software dev consultants/freelancers/contractors to use an agency? If so, what is the benefit?
You don't need a network, you don't need to be a salesman, you don't have to know all the paperwork and you got a guaranteed base salary. At least that was roughly my experience starting out as a consultant, "all" I had to do was show up at clients and deliver well. As an consulting company they come across many opportunities big and small with different requirements, they were quite good at keeping billable hours high at good rates. It was employee-ish but with more variable pay, more varying projects, technologies and people. If you had downtime you could take classes and certifications. Overall a pretty good fast track to build a good CV.
If you mean agencies that take a cut just to match up the work with the people then I think people try to avoid them. Before you needed them to find the resources but with online database it's much easier to find what you're looking for. There's still those who do all that "other" work for a cut though, basically it's more like you buy a support service but if you don't find your own billable work you're not making any money. Often it's worth it to spend time billing and not dealing with everything else. You can waste a lot of time fiddling with paperwork and when you start to put a price on those hours it might make sense. Unless you're the jack-of-all-trades kind that can both sell and deliver and support yourself and be good at all three.
This is the perfect opportunity to move to torrents over darknets such as I2P (but not Tor). This is also the end-state of piracy, unstoppable and untraceable file sharing. It's the last platform switch you'd have to make. The only downside is that it would force media companies to begin an assault on general-purpose computing.
Not worth the effort. The only place they really bother to try holding back the tide are the torrent websites, once you have the magnet link it's such a giant whack-a-mole. The mass lawsuits have gone nowhere, the "six strikes" policy is just fluff, if you want to be the original seeder maybe you need a VPN but... here's 90 million moles to whack. And that's presumably grossly under-reporting the number of private ways people have shared the file. Even if they managed to prosecute and fine 90,000 of those - unlikely - there's like 0.1% chance of that being you. Even if you get hit with the US statutory minimum of $750 for willful infringement - also unlikely - that's like $0.75 risk-adjusted.
In most other countries you can only sue for actual damages + costs and most places an IP address is also not good enough to win a case in a multi-person household, guests, tenants etc. it doesn't stick to the subscriber. So far the advice here in Norway to these "blackmail letters" is to simply ignore them. Don't acknowledge them, don't admit to anything, don't pay anything and they will go away. They just hope that at least some people will basically incriminate themselves or pay to make it go away. That said, a lot of people do pay when there's decent online services.
You're thinking of the incremental advances from current AI. That will certainly be leveraged, but eventually we will come up with general AI in a way which can be accomplished using available resources. That's decades away according to most people, but any country that develops it first can literally out-think the others in everything, unless they don't have enough lead time. Every government in the world would go to war for that power or to keep that power out of the hands of another.
Replace AI with nukes and out-think with out-kill and the rest of the world should have allied and invaded the US in 1945. You're also assuming that a super-intelligence will appear out of nowhere and that country won't build up to a golden age of economic and industrial power on the way. And that said nation won't ally itself with partner states that'll stand in its halo rather than join a conspiracy to dethrone them. For that matter, the assumption that it'll be a nation state is dubious and not some mega-corporation looking for a generic business optimization engine. Oh yeah, you're also assuming a general engine will be better than a specialized engine. The human brain is amazingly flexible, but it's decades since we lost to computers in chess.
The odd part about that ruling is that it established that APIs are copyrighted, but didn't come up with a single plausible theory of how they could possibly be used in a way that is infringing. If you can copy APIs wholesale to create a competing commercial product you're pretty much in the worst corner of on at least two if not three corners of the fair use test. If the functional nature of interfaces demand that they have to look the same to work the same and will grant you a fair use defense every time it's just semantics.
I'll wait until they've done a much larger properly blinded test.
If I had terminal cancer I'd take highly experimental over certain death, you're only risk averse when it usually ends well. When it'll end badly you're ready for any "Hail Mary" save. Not that I'd try obvious snake oil and superstition, but any reasonable experiment I'd be in on... seems like the worst that can happen here is that you get mighty drunk, granted I've had bad hangovers but I'd rather go out drunk as a skunk than wait for the cancer to get me.
The only way around that is to use some technique that trims the blockchain, but I'm not sure if that's even technically possible.
Sure it would, just have the blockchain sign a complete account balance and everyone could restart from there. It could be vulnerable to a 51% attack totally rewriting the balance but you probably also make some sort of rejection mechanism saying that if >1% of the blockchain rejects the account summary you continue from the last universally acknowledged one. And some nodes would probably stay in full history mode anyway. That doesn't seem like a particularly hard problem, compared to all the other ones...
Yeah, but you can eliminate that from the equation because 20/20 vision = one arch-minute (1/60th of a degree) of resolution. So if the TV covers 30 degrees field of vision (FOV), you can see 30*60 = 1800 pixels of resolution (with a perfectly curved screen where you don't have aliasing effects, 3600 if you do). Now most young people have better than 20/20 vision because that's just the threshold we've set for normal, doesn't need correction vision and if you're really enjoying your widescreen you might be pushing 50 degrees though that's uncomfortably close for most. But even if you have a 50 degree FOV and 20/16 vision you'll at most see 50*60*20/16 = 3750 pixels, which is like front row at the cinema with your average teen eyesight.
Exceptional people have all the way down to 20/10 eyesight, though 20/8 is the physical limit. Exceptional people might recognize aliasing, in theory you could arrive at a number where you need 15K resolution (50*60*20/8*2) to display a "perfect" image. If you are a photographer and want to like stare at a detail on the screen ignoring the big picture, a 32K monitor might make sense. But big Hollywood productions like "The Martian" have made a 2.5K master and nobody has complained, truth is even 4K seems overkill for video. It's not like anybody can actually hear the difference between 24 bit / 96 kHz and 16 bit / 44.1 kHz, it's all in their heads. Good for editing but makes zero difference for presentation.
All you are doing is extending the value to other things that also don't have value and saying that is why it has value. Everything in the crypto-currency markets have the same problem. Just because one pump and dump stock is used to trace between other pump and dump stocks doesn't make any one of them valuable.
That's what *all* currencies are about... are there going to be telltale signs of people retreating to "safe" currencies... there's a reason why in times of trouble people invest in USD, swiss francs (I forget the TLA) and EUR it's because they're massive economies that won't fail without an economic Armageddon. Also precious metals like gold and silver that people assume will "always" be valuable without any real reason except for thousands of years of history.
While the price of bitcoin keeps surging people won't use it for things outside of the crypto-currency markets. It must stabilize, and the entire design of it won't allow that. It will either keep surging or it will fall, stable is not what it is designed for.
You're casually dismissing all the actual things bought and sold with bitcoin like they don't exist? Even if it happens to be illegal substances or various other dubious things they're real, as in people who don't give a shit about speculation but who buy bitcoins to buy it. And the dealers sell for "real money" too, as they can sell them to new buyers. They don't care what the going rate is, it's just a proxy. And between speculators and users you have a working currency, whether you deny it or not.
V2E communication is the absolutely least thing needed for autonomous vehicles. Identifying other vehicles, traffic lights and road signs is completely trivial compared to dealing with "everything else". Random people, animals, debris and whatnot will never wear a chip, if one car can't deal with it on its own neither can a group. The potential benefits of "near-psychic" autonomous cars moving as a pseudo-train and timing the crossing of the intersection is nice-to-have to optimize cars that are already self-driving.
A 30% jump in one month, after two decades of "YYYY will be the year of Linux on the desktop!" ? The explanation is obvious: bad data.
Almost certainly, it would take an extreme swing in sales/installations to make the number of active Linux devices swing 30% in a month. After all, the vast majority of people use the same OS this month as they used last month. If you say people make an OS choice every 3 years, you'd at most expect 100%/(12 months/years*3 years) = 2.8% change per month even if everybody switched to a new OS. Also it's not collaborated by other sources, the desktop market share on statcounter has barely moved, Linux has gone from 1.74% to 1.79%. When these curves swing it's because there's been some kind of change in how they identify devices and platforms, either because they changed the definition or because new devices challenge it.
... store everything I've ever written or photographed on one of these. Every personal record, bank statement, tax form. Then sneeze once and its lost in the shag rug forever.
Just ask the NSA for their copy... for bonus point they'll also have all your phone calls transcribed, your photos will be marked with facial recognition and location recognition and cross indexed with the cell phone tower, GPS and open WiFi name records of where you've been and every http URL you've ever visited. And I'm only half joking, it's creepy that technology now actually makes it feasible to store practically everything about practically everyone.
Threadripper de-lidded. Two are filler blanks, two are the 4+4 cores of the 1900X. Idiots and idiot mods, this place has really lost it.
I've always been an AMD fan, but their offerings have been lackluster lately. On my gaming rig I finally went Intel and on my regular desktop machine I'm still running AMD but it's an ancient 7-8 year old Phenom II. I haven't had much reason to upgrade it until now but I'm thinking I may finally pull the trigger on a system built on the "budget" Threadripper.
Note that the budget Threadripper is a 4+4 system with non-uniform memory, if you want it for the PCIe lanes (64), memory bandwidth (quad channel) or memory size (8x16GB = 128GB possible) fine, but it'll have higher memory latency and the idle power consumption is much higher because you have 2 dies + an interconnect that's always on. It'll probably still do slightly better than an 1800X because it clocks higher since the TR platform has higher TDP and beefier cooling but the motherboards are also likely to be more expensive because of all the extra traces.
I think this product is rather niche, most people would be better off with a Ryzen and those who need the TR features often want more cores too. It's a pretty good deal for AMD, they can take two "junk" Ryzen processors with 50% defective cores and sell at higher price/core than a full 1800X with 8 of 8 cores working. If you need the unique features of TR it's still a good deal, but I don't think it's a processor you should casually buy without a rather specific plan as to how you're going to use it.
You forgot the Nissan Leaf, it's been the #1 seller for electric car here in Norway for some time. Though it's definitively been a #2 car for most families, it's not exactly made for road trips. Or any long trip really, it's a commuter car to do you daily chores around town. It works for some as long as you got a family car for the other trips.