I'd argue that a large part depends upon how you distribute the weight and handle your physical connectivity. Twenty points on your upper back is tiring. Ten pounds on your head is worse. Forty pounds anchored to your hips is just an alternative to "leg day" at the gym (or a few hours on a Stairmaster, or cleaning your house and making 200 trips up and down the stairs). Your legs & hips can handle a fair amount of weight without much impact (assuming you're reasonably in shape to begin with).
Likewise, if the power cables are relatively thin and well-integrated into your clothing (or self-tensioned on spring-loaded wheel rolls to minimize slack, that's not a huge issue. People walked around with headphones and Walkman cassette players for 20+ years. Women have carried purses, well, almost forever. Good integration goes a long way towards making wires tolerable.
Combine 20 pounds of hip-mounted battery with one pound of backpack-computer and 8-12 ounces of headgear (optionally tethered wirelessly to a local server on the local LAN and surrounded by a swarm of drones for remote sensing), and you can run for quite a while. Drone battery life is a more serious limiting factor... but if you have enough drones, you can just keep some of them back at the base. As batteries run low, fly the drone back to the base to recharge, and send another one out to take its place. Keep in mind, I'm being "5-20 year visionary" here. Obviously it would be cost-prohibitive to maintain a fleet of two dozen thousand-dollar drones today just so you can keep 6-8 airborne at any given point in time for an entire day everh third Sunday of the month. Build up a bit of a commercial market so you can have a fleet of three dozen drones and rent them out to gamers (so they're in daily use, spreading around your capital costs), and it starts to look sane. Ditto, for VR gear... people who wouldn't spend $10,000 for the "full monty" might happily pay $400 to rent the same gear for the whole weekend (say, LARPers), $250 to rent it for a Saturday or Sunday, $100 to rent it on a weekday or a few hours on a weekend, etc.
Think about it: back in the 80s, those of us who are GenX'ers spent a small fortune (25c at a time) playing videogames on hardware that was too expensive to own for ourselves. By the 2000s, the last of that market kind of dried up (outside of early VR), because the line between "what you can afford if you spend a lot of money" and "what's available to the public on a time-sharing basis" got fuzzier and fuzzier. XR is a brand new market for stuff like this -- too expensive for most people to justify, let alone afford, as individual purchases... but do-able for collective timesharing.
I think there's a case to be made for having a gradient of capabilities and power options.
For example, if you just want environmental metadata (a-la-Google Glass), that doesn't necessarily require a lot of local horsepower (and generally DOES require good, low-latency network connectivity), so cordless/wireless functionality is important and totally do-able.
On the other hand, if you're doing 3D CAD work, or things like you'd commonly see on a TV show like "Agents of S.H.I.E.L.D.", being tethered to both local power AND a beefy mini-supercomputer isn't a big deal, because you really AREN'T going to be getting up and moving around much. Worst-case, maybe use something like the ML Lightpack for local power, recharge the future "Lightpack" with something like Qi inductive charging when seated, and offload the heavy computational lifting to a server (or array of servers) running nearby.
Put another way, it doesn't HAVE to be an "either-or" decision. The correct answer to "Should XR focus on portability or horsepower?" is "yes". If you're playing SimCity XR, being wirelessly tethered to a local server with a few hours of battery life is fine. If you're playing "Paintball XR" by yourself at a park, you'll need extreme portability and fairly intense location/positioning... but can also reduce the amount of processing power needed for environmental mapping by pre-building your environmental database ahead of time, and maybe leaving active waypoint beacons lying around so it doesn't have to work quite as hard to figure out where you are and what (approximately) you're looking at. If you're walking through the mall looking for sales, you can probably live with a single focal point of rendering, and need to do scene-analysis mainly to keep things like AR ads anchored into place on walls, floors, and ceilings so they won't drift around and distract you too badly. The more you can constrain your environment and reduce the amount of realtime data-crunching necessary, the more easily you can get away with having less on-board portable computing power.
The more adhoc realtime unstructured freedom you want, the closer you venture into supercomputer territory (quite probably surrounded by a small army of sensing bots & drones whose job is to scan, survey, and map not only your immediate surroundings, but areas that you HAVEN'T ventured into yet (so the system can be ready when you DO arrive).
Future vision, 5-10 years from now: state or county park somewhere. 5-20 nerdy adult virtual LARPers running around with holographic headgear and haptic body suits, surrounded by a swarm of drones (mapping the environment, tracking both players and NPC-humans within the game environment, acting as airborne 802.11ad access points, and linking to a fairly beefy server running in the trunk of somebody's car). Park visitors not directly involved with the game are treated like "human NPCs" -- virtually re-skinned to look like {whatever}, and tagged by the game as "do not approach". Eventually, people who aren't involved figure out that there's a rule that penalizes players for coming within ~25 feet of a human-NPC, and some park visitors WITHOUT XR gear involve themselves in the game as well by intentionally doing things that force the players to react to them (like intentionally approaching players, who are then forced to run away to avoid coming within 20 feet and being penalized by the game). Think: the perfect fusion of mixed-reality, paintball, LARPing, and live-action Quake. Good times:-)
> There are no ways that we know of to radically increase computer power.
Of COURSE there are. They just aren't cheap, low-power, or within the capabilities of semi-passive cooling.
Do you REALLY believe that an agglomeration of i9-class CPUs (in good old fashioned SMP, not performance-compromised multicore chips) with water cooling & the kind of heat output we accepted as normal in the Pentium 4 era can't radically raise the performance bar?
We've settled for stagnated performance in a race to cheaper + thinner + lower-power for years because nobody could give consumers a reason NOT to settle for it. We need to raise the bar, raise it HIGH, and give people a compelling reason again to CARE about performance and demand it. We need to give Joe Sixpack a REASON to go spend $5,000 on a personal supercomputer that would have made 1999 NASA jealous.
I don't remember who first said it, but mixed reality is really just augmented reality with better, evolved display tech. It's not an end in itself. AR display tech is a cornerstone, but it's just one piece of the equation.
The ML1 might not be the lightfield display we've fantasized about (yeah, me too), but it's the best holographic display you can buy RIGHT NOW. The technology will unquestionably keep improving... but you have to start somewhere.
We didn't magically get to photorealistic 144fps 3840x2560 videogames overnight, and I have a mountain of old hardware in my closets to prove just how far we've come. Enjoy the ML1 for what it is, and look forward to what it's paving the way towards next... and pray that the race to 'cheap' doesn't win out over the quest for 'awesome' before XR technology GETS to awesome.
Part of the reason computing power stalled was due to hitting die-shrinkage limits, but much of it is due to the expectation that modern hardware has to be dirt cheap. I paid around $2500 for an a Amiga 3000 circa 1992. What kind of a beast of raw, brute-force AMD64 power could you build TODAY with that same inflation-adjusted amount (say, ~$7,000)? If people still routinely bought $2,500-3,200 laptops & tolerated inch-thick 10lb form-factors, what kind of laptop power COULD we have now?
Thin, light, and power-sipping has won for now... and the computer industry needs a reason to say, "fuck all three, give me a backpack-sized mainframe". AR *is* that reason.
The ML1 isn't cheap, but its price really isn't out of line with what you'd spend if you bought an Oculus Rift and a gaming laptop comparable to what's inside the ML1's LightPack controller.
If anything, the single biggest problem with mixed/augmented/virtual reality today is that it really needs way more horsepower than any mainstream (let alone cheap) consumer device currently has. Current hardware is kind of like a NeXT back when it was the computer to die for... lots of promise & future-looking software, running on hardware that just wasn't quite fast enough to satisfy people's expectations.
In all honesty, XR (my favorite umbrella term for mixed/augmented/virtual reality) is what the currently-moribund PC industry NEEDS... an excuse to RADICALLY increase computing power. We haven't had an excuse like that for 10 years. The same beefed-up hardware that will enable realtime XR applications with low latency and fluid animation will finally give us things like "Aero Diamond" (Aero-like Windows graphics, but with realtime-raytraced eyecandy and translucency effects) once even a mid-range laptop has the equivalent of today's most expensive hardware.
NVidia has taken the next step towards realtime hardware-accelerated raytracing, and Intel & AMD have started moving into 8+ core 5+GHz territory. Pair the display hardware of a ML1 or Hololens with a 16-core i9 running at 4.5-5GHz with 64gb of RAM, a 2TB SSD, and a top of the line dual-slot NVidia GPU (call it "personal cloud"), and watch the real magic happen. Pair the same display tech with the equivalent of a high-end Android phone, and prepare to be kind of underwhelmed, just like we were 25 years ago with NeXTSTEP. The fundamental idea is good, it just needs radically more-powerful computer hardware driving it to make it truly awesome.
The only people who think replacing keyboards for general text input is a good idea are people who never bothered to learn how to type.
Mobile devices might be an exception due to size... I'm a Graffiti person myself (originally Palm, now Android). It's how I'm able to make long postings like this one with my phone. Predictive keyboards slow me down, because I'm kind of OCD about not making spelling errors & only Graffiti is accurate enough to let me get away with NOT proofreading everything word-by-word. My only gripe with Android Graffiti has to do with capacitative behavior & governor-induced lag... it pisses me off that a fsck'ing sub-20Mhz 680x0 could handle nearly error-free input, but an Android phone with a faster CPU & more cores occasionally can't tell the difference between 'o' and 'u' because... er... Android decided something BESIDES high-res stroke-capture is a higher priority at that moment (using a governor like 'performance' or 'responsive' on a rooted phone definitely helps).
That said, there's room for improvement with Qwerty. The patents on the Matias Halfkeyboard expired years ago... it should now be a default feature on any new keyboard (and Matias himself should sell a USB adapter to sit inline and turn ANY USB keyboard into one), not to mention a standard feature in Linux and Windows. It's handy for "left hand on keyboard, right hand on mouse" use cases.
At this point, the main point of something like the ML1 isn't just "rendering floating metadata bubbles next to interesting objects", it's "being able to visualize things in 3D, without many of the drawbacks that severely limit the amount of immersive virtual reality someone can take before they start getting queasy."
The fundamental problem with fully-immersive virtual reality is that current VR hardware just isn't fast or responsive enough to sail over the uncanny valley. If you're wearing a ML1 and rendered objects quiver a bit as you look around, your brain interprets it as "hey, there's motion over there... you should probably pay attention to it, just in case it's a tiger hiding behind a log". If you're wearing an Oculus Rift and the entire virtual room around you sloshes, lags, and jerks around, your brain just goes into meltdown because observed reality is now in frank contradiction of everything your OTHER senses are telling it. And if you're wearing a Rift, looking at some 3D virtual object that follows your head motions, but otherwise remains fixed & motionless regardless of YOUR OWN motion, you'll end up with good old-fashioned motion sickness.
Put another way, mixed reality buys you a metaphorical hall pass & loophole to get around your other senses, by ensuring that MOST of what you see is in agreement with your other senses, regardless of whether specific sub-details accounting for a small percentage of your surroundings are in agreement as well. The more your larger environment appears to be consistent with your other senses, the less likely you are to be affected by vertigo.
You have to sign up for access, and the number of hoops you have to jump through to get into the site for the first time and see anything at all is kind of annoying... but once you're in, the site is quite good, and generally fluff/bullshit-free. It's almost like the development team hid it to keep it safe from the marketing team's interference. It's a total night/day difference from the "main" site... lots of real, meaty content. Anyone can register, and it's definitely worth the effort if you're even slightly curious.
Big tip: make a special effort to recursively work your way down the "Bootcamp in a box" section... it links to other important sections that aren't necessarily obvious or easy to find from the menu on the left (that was 1-2 months ago... it might have improved since then).
When running the various "Hello Cube" variants for the first time, remember that the real device has a limited field of view where holographic content can go, and the simulator faithfully emulates this behavior. If you don't see the cube, look around the whole virtual room VERY carefully and systematically, sweeping through a 360-degree arc approximately between the floor and ceiling. Chances are, the cube IS there somewhere... it might even be located within the part of the room you can see, but outside the zone where holographic content can be.
You also might be too close to the cube to see it -- anything surface closer than ~18 inches gets culled by Unity, and remember, the INSIDE/REAR surfaces of Unity/OpenGL triangles are normally culled & invisible if you're looking at them as well, so if your virtual nose is metaphorically poking into the cube, you probably won't see ANYTHING at all. If you don't see anything after exhaustively & systematically looking around the room in a 360-degree circle, try stepping back ~4 feet and look around the room again.
Pay special attention to the "Planes" and "Raycast" examples. If your biggest question after a couple of days is, "How do I make sure {something} ends up on the table, or sofa, or a specific area of the floor", it's probably because you haven't worked through the Planes & Raycast examples yet, and as a result you're missing a fairly fundamental point of mixed-reality development. With early augmented-reality development for devices like Google Glass, you could skirt around the problem because conceptually, everything you rendered was being displayed on a ghostly planar surface a few feet away. Magic Leap takes it up a few notches, and allows you to choose the apparent distance of an object, too. With freedom and power comes responsibility... being able to place objects semi-arbitrarily within a room means that at some level, your program has to be at least vaguely aware of what's IN the room surrounding you, and take that into consideration when deciding where to put things. You CAN'T just blindly position objects at arbitrary coordinates & expect everything to "just work"... you have to make sure that you're either putting things where nothing already exists in the real world, or be ready to deal with the consequences OF making ghostly holograms coexist within the same volume of space as a real-world object.
Also, keep in mind that the ML1 semi-passively builds its internal model of a room in the background as you look around. This also applies to the simulator. If something involving planes/raycasting to find surfaces doesn't seem to be finding a surface that you KNOW is nearby, help it out... slowly look up and down around the room, and do it from several vantage points. The more data you give it (by looking around the room in a systematic manner, from different angles), the better its model of the room will be.
Update: I was wrong about maritime law & salvage rights. Apparently, the US isn't a signatory to the UN treaties governing maritime salvage, and takes the position that a) sunken US government ships that sink on the high seas remain the exclusive property of the US in perpetuity and are never implicitly abandoned; b) salvage rights to ships that sink in US waters remains with the original owners in perpetuity... and solely belongs to the US government if the wreck gets salvaged by anyone else; and c) jurisdiction over manmade objects in space remains exclusively with the nation-state associated with its launch in perpetuity.
So even IF Hubble were in uncontrolled freefall, if SpaceX fixed Hubble on its own, NASA would (at best) say, "Thanks for fixing Hubble at your expsnse... it's still ours." And SpaceX & its employees *could* conceivably be prosecuted on federal charges for touching Hubble without permission if the feds were determined to show the world that no good deed ever goes unpunished.
TL/DR: The US explicitly rejects most modern international maritime law, and in any case asserts that maritime law does NOT extend to space.
A Trackpoint isn't a touchpad... it's a tiny isometric pointer stick commonly associated with Thinkpads. And the whole point is that it enables you to use the mouse without having to lift your hand from the keyboard (something that the finger-pointing unwashed masses just don't seem to grasp the importance of).
I hate finger-pointers. They're the users who ruined touchpads. Back in "the old days", touchpads acted like flat trackballs, but adjusted curving gestures to appear straight like a thumb-trackball did. Then, manufacturers all jumped on the "optimize touchpads for finger-pointers instead of thumb-sweepers", and touchpads became unusable by thumb-sweepers. They COULD have made touchpads dual-mode, but instead they just eliminated the thumb-optimizing DSP logic entirely. Fuckers, that's why we can't have nice things...
I personally don't get the obsession with eliminating the top, and especially the bottom, bezels. Width affects the max screen size that can fit in a back pocket, but portrait-orientation height? Pffft.
Persondlly, I *like* having a reasonable bottom bezel, because it's really hard to hold a large, heavy phone with one hand & operate the back/home/whatever buttons with my thumb if they're all grazing the lower edge. IMHO, the most grippable phone I ever owned was my old Galaxy S3 with Seidio extended battery case... it put the center of mass comfortably low & provided a handy ridge to grip the phone with from behind. The phone couldn't lie flat, but it had a kickstand which mitigated 99% of the annoyance & overall, it rocked. Even though I eventually bought bigger batteries, I kept going back to the Seidio battery/case because it was SO much easier & more comfortable to hold.
The big question, though... if you keep the phone in an Otterbox Defender-like case & the rear glass cracks, do you HAVE to fix it, or can you just say "fuck it, it's hidden inside the case anyway" & just leave it cracked without affecting anything besides "out of case" aesthetics?
Also, does the glass help or hurt with heat-removal? Can the phone run at 100% max from power-on until battery-drained (say, if you're using Daydream VR), or will it go into thermal-throttling within a few minutes if you try (like a Nexus 6P does)?
We might not eat 100% of a cow, pig, etc... but the net waste is practically zero. AFAIK, practically every cell of a cow now has commercial value for something. A friend who once had a summer job at a slaughterhouse told me that most of the trash leaving a slaughterhouse comes from garbage cans in the offices & break room, and most of the remainder comes from the janitorial or maintenance departments (detergent containers, old knives, etc) & ends up getting recycled.
If you want to try out a fun product that will whet your appetite for a Magic Leap ML1 or Hololens, check out Aryzon's headset (http://aryzon.com). It's basically the augmented-reality evolution of Google Cardboard -- you mount your phone in it, the display's light bounces from a mirror onto reflective window tinting, and you see a 3D hologram superimposed on whatever is in the room in front of you.
Aryzon's lead (only?) developer has done a fantastic job of writing documentation, creating demo apps, and making tutorial videos ( https://www.youtube.com/channe... ).
Make no mistake... it's nowhere close to being in the same league as a ML1 or a Hololens... but at 30 Euros (~$36) with free shipping worldwide (including to the US), it's actually cheap enough to buy for shits & giggles. Don't be afraid to order one just because it involves mailing to the US from the Netherlands... NL-US mail is pretty fast (it also might be available from Amazon, if you found this post via Google long after it was a current topic on Slashdot).
There are two similar products I'm aware of to Aryzon's headset -- HoloKit and the Lenovo Lightsaber:
* Holokit's website (holokit.io) has been dead for at least the past few days, and probably longer. You can get it from Amazon with 2-day Prime shipping... but I'm not really sure what you could DO with it once it arrives, because as noted... their website is missing in action. And AFAIK, HoloKit doesn't have a headstrap, so you have to use one or both hands just holding it up. Speaking as someone who's done the whole "hold a Cardboard viewer with a heavy phone on your face using one hand" routine in the past, I can definitely say that it gets old (and tiring) really fast.
* The Lightsaber is probably a step-up design-wise, but (AFAIK) has no public SDK. Lenovo has apparently been promising one since last year. Enough said. Without a SDK, it's just a silly toy.
Aryzon is a small company, but their main (only?) developer has done an amazing job of writing demo software, creating documentation for developers, and posting tutorials on Youtube (https://www.youtube.com/channel/UCy7KJ7TQ3myJ0fSVqFDudeg/videos).
That said... MagicLeap and Hololens are more than just polished hardware. They both have a substantial amount of software backing them up to do things like 3D surface-mapping and immersive binaural sound. Do you need it if your entire goal is to admire a holographic balloon floating a few feet in front of you? No, not really. But you're really, really going to need it if you want to render something onto a random tabletop, floor, wall, or ceiling in a random room somewhere... because THAT'S a really HARD problem to solve. So hard, in fact, that when you first get into mixed/augmented-reality development, you probably won't even realize it IS a problem until you've been at it for a few days... and then, the sheer enormity of it hits you like a boulder.
TLDR: if $2,295 is out of your budget, buy an Aryzon. It's cheap & fun, and it'll give you an affordable taste of why holographic mixed reality is so cool. Just remember that there IS more to "mixed reality" than "being able to see digital holograms superimposed on whatever's in front of you", the same way there's a world of difference between experiencing 6DoF immersive VR with an Oculus Rift, haptic gloves, and full-body motion tracking and experiencing the equivalent of a Viewmaster where you can stand fixed in one spot and enjoy a 360-degree view by moving your head around.
Better analogy: you own a big, huge, old London-style double-decker tourist bus on a small, remote island. Say, somewhere in the South Pacific. Twenty five years ago, the bus was delivered to the island by a unique amphibious cargo ship capable of surviving the high seas, rolling up onto the beach, and driving it off before departing to haul big things to other remote islands.
The bus is now old and decrepit... but amphibious ships like the one used to deliver your bus no longer exist... they had some really bad accidents (partly due to bad management decisions), were declared to be unsafe, and quickly retired with a huge backlog of scheduled deliveries that are now impossible to make. On the bright side, the island now has twice-daily low-cost jet service to the mainland... great if you're a passenger or need to ship a small package, little consolation if what you REALLY need to ship is a big, huge double-decker bus that won't FIT inside a modern jet.
It's even worse than that. JWST isn't a real REPLACEMENT for Hubble. It'll be able to do things that Hubble can't, but there are even MORE things Hubble can do that JWST won't ever be able to do.
The fact is, we don't have the ability to launch a new satellite as big or as heavy as Hubble... and AFAIK, there are no rockets even under development that will be capable of delivering something as physically BIG as Hubble (or the existing ISS modules) into orbit. Hubble and the existing ISS modules are all we have, and all we're LIKELY to have for DECADES. They're literally irreplaceable within the span of our lives, and as such, deorbiting them is, IMHO, wantonly reckless and irresponsible. If Hubble's telescope functionality dies before we have the ability to send a refurbishment mission, we should be ready to DO the deorbiting mission on 18 months' notice... but always and only as a last-ditch "plan B" if we don't get the ability to launch a robotic or manned servicing & refurbishment mission first.
From what I understand, if Hubble failed completely and became totally uncontrolled tomorrow, it would be at least a decade before it fell far enough to present imminent risk of uncontrolled reentry. If we made even a token attempt to send a robotic mission to boost it into a higher orbit, we could easily add another decade to that. SpaceX might not be ready to fly a refurbishment mission to Hubble within the next 3-5 years... but it probably COULD be ready to do it 6-10 years from now (if it had a firm commitment from NASA), and will probably be capable of doing it within 10-15 years regardless of what NASA does (and knowing Musk, would probably invoke maritime salvage law & refurbish Hubble ITSELF as a commercial venture if NASA couldn't/wouldn't do it).
Frankly, I think 90% of NASA's sense of deorbiting-urgency is precisely BECAUSE they'd rather see things like the ISS and Hubble get intentionally destroyed than risk allowing someone else to metaphorically grab them from the curb before the garbage truck arrives.
I just want to see the FCC adopt more flexible rules governing spread spectrum, esp. with regard to HF and low-bitrate QRP digital modes.
I think it was Bruce who argued a few years ago that EIRP per hz per time-interval (within a larger umbrella applying to instantaneous power & bandwidth) matters more than the name and legal definition of any particular mode.
If anything, the test for Amateur Extra should include things relevant to advanced digital modes, like OFDM, CDMA, FFT calculation, SDR, matrix math, etc.
Maybe set aside a 100MHz sandbox for the lowest license-class hams to play with without restriction (besides input power and total bandwidth) at low power in the 10GHz band (almost nothing you can do with a watt at 10GHz is going to matter much, even if you screw up in nearly every conceivable way), expanding bands & power as you move up the license classes (reserving HF wideband spread-spectrum for amateur-extra class (which is something that WOULD motivate me to move up from General class).
The idea being that someone who makes it to AE will hopefully at least understand HOW things can innocently go wrong with exotic digital modes, so the FCC can give them more freedom to play in situations where allowing a (novice-)Technician class licensee would be like handing a loaded gun to a toddler.
And how, exactly, would compromising a 2m/70cm radio with a range of a few miles likely to be of any use to the Chinese government? It transmits FM audio. ANYONE with an antenna & receiver nearby can receive it, by intent of the radio's owner.
And even IF it could somehow be hijacked by Chinese agents... the last time I checked, China's government has no legal jurisdiction over me (an American citizen in America), and I'm not going to flatter myself into thinking that anything I might do is of the SLIGHTEST interest to them.
There's awareness, there's paranoia, and then there's silliness. If you're going to lose sleep over a $29 pocket radio from China, I sincerely hope you don't own an iPhone, most Android devices, ~97% of the LCD TVs in existence, or anything that combines networking and a microphone. Worrying about the security implications of using a Baofeng radio is like losing sleep about sea-level rise because it rained yesterday.
In the US, from the perspective of common law, the buyer would technically have no recourse in a purely civil claim unless he could establish that he suffered harm that can be quantified in dollars... and that he actually DID suffer a net loss. So, he might still have an actionable claim if he suffered indirect harm whose cash value exceeded the instant profit from the art doubling in value... but it would be an uphill battle to quantify the cash value of an emotion like "disappointment".
There's also & independently the possibility of criminal prosecution ("conversion", for example, might be a tort... but in most jurisdictions, it can ALSO be a criminal offense under certain conditions). The catch with the criminal angle is that the buyer purchased "experimental art" from someone KNOWN for doing things like this, so Banksy could semi-legitimately argue that the unanticipated public shredding was itself integral the art & what was purchased.
TL/DR: legally, it's the kind of case only the lawyers would collectively win. If the buyer is unhappy, their best recourse would honestly be to just flip the painting to someone else & walk away. At most, a vengeful buyer might spend a fortune, waste time testifying in court, give Banksy a metaphorical black eye... and be blacklisted as a potential art buyer going forward. "Experimental art" is, by definition, kind of fucked up by "normal person standards", and that applies to BOTH artists AND patrons.
Actually, I just remembered... dealing with symlinks in Windows is still kind of a mess. A few years ago, I tried writing a Java program that would write a file, then create (or replace) a symlink in another directory so it always pointed at the latest file.
It turned out to be hopeless. Not because Java can't create or modify symlinks (it can), but because Windows has bizarre, fucked-up rules governing who is (and more importantly, isn't) allowed to create & modify symlinks. From what I remember, one of the rules is that normal users can create and modify symlinks (in directories they own), but NOT users who are ALSO members of an 'admin' group, in which case theyre only allowed to do it if the program is running with elevated permissions. For the life of me, I fail to understand why it's ok for a user who's one step above "guest user" to create & modify symlinks in any directory he owns, but taboo for a user who's ALSO an admin -- but not running AS an admin -- to do the exact same goddamn thing in his own "Documents" folder.
I could see Windows enforcing a rule like that if you want to create or modify symlinks in a system directory (or even a directory like "C:\Program Files"), but extending the prohibition to ANY directory SYSTEMWIDE (including the user's OWN documents folder) is just plain fucking stupid.
Windows is littered with stupid rules like that. I remember a few years ago when a (seemingly) minor Windows update instantly broke every C# Microsoft Office extension we had, because Explorer and NTFS both enforce different (and not necessarily consistent) rules, and Office enforces yet another set of rules that's consistent with neither Explorer nor NTFS, and partially depends upon the syntax you use to specify paths. From what I recall, we were trying to read and write to files in a subdirectory of a network share that was mapped to a drive letter... previously, it was OK, but under the new rule, we had to use extended-UNC notation. Only in Microsoft's wacky bizarro world is it a permissions error to try and read "z:\path\to\taboo.dat", but totally OK to do the same thing to "\\?\UNC\somehost\foo\path\to\taboo.dat" when both refer to the EXACT SAME GODDAMN FILE.
The point everyone seems to miss amidst the moral panic is that the risk from consuming an artificially-sweetened 2L bottle of soda might not be ZERO, but it's still a huge net improvement over the known, documented harm likely to arise from the daily consumption of a 2L bottle of soda sweetened with sugar or HFCS.
Daily consumption of 2L of diet soda: theoretical changes in gut bacteria.
Daily consumption of 2L of regular soda: significantly elevated risk of diabetes and obesity.
The worst thing that could possibly happen is if someone who's already slightly obese, who drinks 2L/day of diet soda, were to read this and say "fuck it, it's healthier to drink regular soda instead." Because it's not.
I think it was partly pollution, but also partly the realization that if she hadn't, 99% of the trees within 50 miles of London would have been gone within a century. Wood just doesn't have the energy-density necessary to provide winter heat to big cities with hundreds of thousands of residents... not only does it require lots of time and land to grow, but the logistics of delivering it (in sufficient quantities, with sufficient frequency) become insurmountable as well.
I'm pretty sure that's at least part of the reason why in 1400, cities like Rome, Athens, and Constantinople/Istanbul had been large, sprawling cities for centuries, but cities like London & Paris were still overgrown forts & trading posts. At least in cities like Rome, heating a dwelling was more of a luxury and matter of personal comfort than a matter of literal life and death. It's one thing to be unable to heat your apartment when it's 40 degrees outside at night for a few days per year... it's another matter ENTIRELY when it's 20 BELOW ZERO outside at night, and below freezing during the day, for weeks at a time (eg, London). Cities like London didn't have the luxury of being able to treat heating like... well, a luxury. So it became a factor limiting the city's ability to grow.
1. Bend over backwards to configure Windows to ask your "permission" before installing updates,
2. Aren't unlucky enough to have Windows throw up a sudden, "Windows wants to install updates... [REBOOT NOW!] learn more" (with "(learn more)" neither appearing to be an obviously-clickable button nor underlined link, printed in a tiny font, and probably inserted into the middle of a longer sentence that itself is neither clickable nor calls obvious attention to itself... and REBOOT NOW! being the default choice that gets selected if you aren't looking at the screen,didn't notice that Windows has rudely grabbed input focus, and press the 'enter' key).
3. Have been making continuous backups (killing much of the performance benefit to having a fast SSD, unless you ALSO have a fast SSD to use for backups as well)
4. Windows doesn't decide to fuck with your backup drive, too (which has to be continuously connected if you want to satisfy requirement #3).
5. You're ALSO doing backups to an external drive that gets physically removed from the computer and locked away, so it won't be affected by ransomware, or get stolen/destroyed as well if you get robbed or your house burns down).
My biggest beef with pretty much every current Windows backup "solution" is the fact that NONE of the ones that are remotely affordable can gracefully deal with what I call, the size/importance gradient & properly juggle multiple strategies... say, doing frequent incremental backups to a connected drive when the computer is genuinely inactive (say, screensaver active), replicating those local backups to a networked drive on the local LAN when the internal-to-internal backups are finished & the computer is still inactive, and periodically backing up a subset of files from the networked lan backup to "the cloud" (the ones that are important enough to pay ongoing fees to safeguard... I might have ~10-12 terabytes of files, but only a few hundred gigabytes of them are really what I'd be devastated to lose, and only a few gigabytes of them are truly what I'd call irreplaceable... and at least a quarter of THOSE are sitting on remote git repositories somewhere, anyway). Put another way, indiscriminately and continuously keeping 10+ terabytes of files backed up in "the cloud" just isn't sane (in terms of cost OR ongoing performance), but keeping a subset of them appropriately backed up is a major pain.
What I'd LOVE to see from Microsoft (and what would convince me to switch to Linux once and for all, if it appeared there first) -- a new kind of filesystem I'll call "DHFS" ("data-hoarders' file system"). In day to day use, it would work like ext4 or NTFS... files get written directly to the drive, for performance. However, in the background (as a lowest-priority system task), it goes through periodically, finds all the "real" files, moves the "real" file to some "master" volume (assuming there isn't already an authoritative, identical copy of that exact file already there), and replaces it with a DHFSymlink (which works like a "regular" symlink for reading purposes, but writes/updates result in the symlink getting blown away & replaced by the new data (or if it's an append, some special structure that says, "the start of this file is (some file pointed to by this link), but the subsequent data is (this)", since creating an entire new copy of the file just to add a byte to it would totally kill performance).
With something like DHFS, only the authoritative master copies of the files, and their MUCH smaller symlink data, gets backed up.
In effect, DHFS would systematically preserve directly-usable copies of every unique file on the system (assuming you could figure out the "real" name and location of any given file), while preserving the CONTEXT & Metadata of those files with something similar to symlinks. If the ability to roll back state wasn't required, it would also do something akin to mark & sweep garbage collection... it would periodically go through its ar
One big problem I see... as far as I can tell, the standard does nothing to clarify whether a given device supports ONLY 2.4GHz, or whether it supports 2.4GHz *and* 5GHz... and if it supports 5GHz, which channels it supports & how.
There are lots of 802.11ac devices, for example, that either don't fully support the use of U-NII(2C) channels, or have crippled DFS implementations that use a sledgehammer instead of a scalpel... satisfying the FCC's requirements, but doing it in a way that results in a product so crippled it almost might as well not even bother with U-NII(2C) channels(*).
I'm also curious to know how they intend to deal with things like AP-roaming and dynamic handoffs... something that was theoretically defined on paper way back in 2008, but (AFAIK) has NEVER really worked properly with consumer devices on home networks. Or pretty much ANYTHING besides a tightly-controlled Enterprise network.
This is my major beef with wireless network gear today... it's DAMN NEAR IMPOSSIBLE, even if you know EXACTLY what standards you need compliance with, to actually walk into a store like Best Buy and make an informed purchasing decision based on their advertising literature and packaging. And if you DO go online and read teardown reviews, there's still no guarantee... the manufacturer could have completely changed not only the general design, but literally changed out the entire chipset with a completely different one that has inferior performance or standards support while keeping the model number (often, even the UPC) unchanged. Linksys & Netgear are both notorious for this... often, they'll indicate the revision on a sticker on the device itself, but put NOTHING on the packaging that's visible before you break the shrinkwrap to indicate whether you're getting the one that earned 5-star reviews & had people drooling, or the later version that got 1 & 2-star reviews and is a pale, cruel imitation of its earlier self.
----
(*) Many of the 5GHz channels share spectrum with weather radar & have to make a "robust" attempt to detect its presence and refrain from using frequencies where it's detected. The variability comes from the fact that some devices take the cheap approach... shutting down and going dark entirely for at least a minute to listen for radar, and assuming the worst at the slightest hint of a signal. This makes the FCC happy, but results in a product that's dysfunctional (to put it nicely).
The cheap/usual way is for the AP to just "go dark" for a minute while listening on the channel for things like weather radar transmissions. When this happens, the wifi connection appears to just silently drop for no apparent reason, then reappears about a minute later (assuming it didn't detect what it thought was a radar signature).
The more sophisticated way is for an internet-connected device to include a GPS receiver & query the FCC's database directly after discerning its location... if it's not within range of a known radar site, it can skip the majority of physical DFS checks requiring radio silence (basically, doing it once at startup). Alternatively, the AP could include an additional radio receiver & logic so that prior to "going dark" on a DFS channel to listen, it could temporarily switch to an alternate, non-restricted 5GHz channel (and notify clients it's about to change channels). The alternate channel might suck, but IMHO, "works poorly for a minute " is STILL a huge improvement over "goes dark and doesn't work AT ALL for a minute every hour"
I might be wrong, but I think 802.11ad MIGHT (in theory) have the necessary hardware to simultaneously use two non-contiguous 5GHz channels (say, 36 and 104), in which case it could (conceivably) "go dark" on channel 104 for DFS while maintaining an active link with connected clients (at half-throughput) on channel 36... but whether any 802.11ad device you can actually go out and purchase TODAY as a non-Enterprise customer (or any random electronic device conceivably purchased at Best Buy or Walmart) can actually SUPPORT that is anybody's guess.
A message warning about an earthquake arriving in 4 seconds still has merit... if nothing else, it increases the likelihood that your phone (and for women, your purse) will be in your hand (rather than 7 feet away) when the earthquake strikes, maximizing your likelihood of being able to use it to call for help if you get partially buried under debris (or at least ensuring that you're ready to grab the phone and/or purse a moment later, instead of having to stop & think about it).
> The GSM protocol includes a broadcasting feature that overrides all other transmission > in order to deliver emergency messages to all cellphones simultaneously. > Why then the delays?
Because roughly half the cell phones in America have historically been CDMA devices, not GSM. CDMA generally had comparable functionality (on paper, at least), but wasn't literally identical.
Compounding the problem, major parts of CDMA's functionality was officially "optional" & left up to the carrier to pick & choose. Qualcomm intentionally allowed Sprint & Verizon to implement CDMA in slightly-incompatible ways... and Sprint & Verizon liked that, because it meant that even a theoretically-unlocked phone from one network would be forever crippled & dysfunctional on the other, EVEN IF a user managed to get it activated somehow.
Ultimately, it was (mostly) Apple & Google who put an end to much of the silliness. Blackberry & Sidekick mitigated it... but only for THEIR devices... and used their mitigations as a way to try and lock out Palm & Microsoft. The main thing that saved Apple & Google was Microsoft's purchase of Danger & subsequent willingness to license out their patent pool on fair & non-discriminatory terms (and why Microsoft makes more in profits from the sale of an iPhone or Android phone than it ever did from the sale of a Windows Mobile phone).
Slashdot Mirror
I'd argue that a large part depends upon how you distribute the weight and handle your physical connectivity. Twenty points on your upper back is tiring. Ten pounds on your head is worse. Forty pounds anchored to your hips is just an alternative to "leg day" at the gym (or a few hours on a Stairmaster, or cleaning your house and making 200 trips up and down the stairs). Your legs & hips can handle a fair amount of weight without much impact (assuming you're reasonably in shape to begin with).
Likewise, if the power cables are relatively thin and well-integrated into your clothing (or self-tensioned on spring-loaded wheel rolls to minimize slack, that's not a huge issue. People walked around with headphones and Walkman cassette players for 20+ years. Women have carried purses, well, almost forever. Good integration goes a long way towards making wires tolerable.
Combine 20 pounds of hip-mounted battery with one pound of backpack-computer and 8-12 ounces of headgear (optionally tethered wirelessly to a local server on the local LAN and surrounded by a swarm of drones for remote sensing), and you can run for quite a while. Drone battery life is a more serious limiting factor... but if you have enough drones, you can just keep some of them back at the base. As batteries run low, fly the drone back to the base to recharge, and send another one out to take its place. Keep in mind, I'm being "5-20 year visionary" here. Obviously it would be cost-prohibitive to maintain a fleet of two dozen thousand-dollar drones today just so you can keep 6-8 airborne at any given point in time for an entire day everh third Sunday of the month. Build up a bit of a commercial market so you can have a fleet of three dozen drones and rent them out to gamers (so they're in daily use, spreading around your capital costs), and it starts to look sane. Ditto, for VR gear... people who wouldn't spend $10,000 for the "full monty" might happily pay $400 to rent the same gear for the whole weekend (say, LARPers), $250 to rent it for a Saturday or Sunday, $100 to rent it on a weekday or a few hours on a weekend, etc.
Think about it: back in the 80s, those of us who are GenX'ers spent a small fortune (25c at a time) playing videogames on hardware that was too expensive to own for ourselves. By the 2000s, the last of that market kind of dried up (outside of early VR), because the line between "what you can afford if you spend a lot of money" and "what's available to the public on a time-sharing basis" got fuzzier and fuzzier. XR is a brand new market for stuff like this -- too expensive for most people to justify, let alone afford, as individual purchases... but do-able for collective timesharing.
I think there's a case to be made for having a gradient of capabilities and power options.
For example, if you just want environmental metadata (a-la-Google Glass), that doesn't necessarily require a lot of local horsepower (and generally DOES require good, low-latency network connectivity), so cordless/wireless functionality is important and totally do-able.
On the other hand, if you're doing 3D CAD work, or things like you'd commonly see on a TV show like "Agents of S.H.I.E.L.D.", being tethered to both local power AND a beefy mini-supercomputer isn't a big deal, because you really AREN'T going to be getting up and moving around much. Worst-case, maybe use something like the ML Lightpack for local power, recharge the future "Lightpack" with something like Qi inductive charging when seated, and offload the heavy computational lifting to a server (or array of servers) running nearby.
Put another way, it doesn't HAVE to be an "either-or" decision. The correct answer to "Should XR focus on portability or horsepower?" is "yes". If you're playing SimCity XR, being wirelessly tethered to a local server with a few hours of battery life is fine. If you're playing "Paintball XR" by yourself at a park, you'll need extreme portability and fairly intense location/positioning... but can also reduce the amount of processing power needed for environmental mapping by pre-building your environmental database ahead of time, and maybe leaving active waypoint beacons lying around so it doesn't have to work quite as hard to figure out where you are and what (approximately) you're looking at. If you're walking through the mall looking for sales, you can probably live with a single focal point of rendering, and need to do scene-analysis mainly to keep things like AR ads anchored into place on walls, floors, and ceilings so they won't drift around and distract you too badly. The more you can constrain your environment and reduce the amount of realtime data-crunching necessary, the more easily you can get away with having less on-board portable computing power.
The more adhoc realtime unstructured freedom you want, the closer you venture into supercomputer territory (quite probably surrounded by a small army of sensing bots & drones whose job is to scan, survey, and map not only your immediate surroundings, but areas that you HAVEN'T ventured into yet (so the system can be ready when you DO arrive).
Future vision, 5-10 years from now: state or county park somewhere. 5-20 nerdy adult virtual LARPers running around with holographic headgear and haptic body suits, surrounded by a swarm of drones (mapping the environment, tracking both players and NPC-humans within the game environment, acting as airborne 802.11ad access points, and linking to a fairly beefy server running in the trunk of somebody's car). Park visitors not directly involved with the game are treated like "human NPCs" -- virtually re-skinned to look like {whatever}, and tagged by the game as "do not approach". Eventually, people who aren't involved figure out that there's a rule that penalizes players for coming within ~25 feet of a human-NPC, and some park visitors WITHOUT XR gear involve themselves in the game as well by intentionally doing things that force the players to react to them (like intentionally approaching players, who are then forced to run away to avoid coming within 20 feet and being penalized by the game). Think: the perfect fusion of mixed-reality, paintball, LARPing, and live-action Quake. Good times :-)
> There are no ways that we know of to radically increase computer power.
Of COURSE there are. They just aren't cheap, low-power, or within the capabilities of semi-passive cooling.
Do you REALLY believe that an agglomeration of i9-class CPUs (in good old fashioned SMP, not performance-compromised multicore chips) with water cooling & the kind of heat output we accepted as normal in the Pentium 4 era can't radically raise the performance bar?
We've settled for stagnated performance in a race to cheaper + thinner + lower-power for years because nobody could give consumers a reason NOT to settle for it. We need to raise the bar, raise it HIGH, and give people a compelling reason again to CARE about performance and demand it. We need to give Joe Sixpack a REASON to go spend $5,000 on a personal supercomputer that would have made 1999 NASA jealous.
I don't remember who first said it, but mixed reality is really just augmented reality with better, evolved display tech. It's not an end in itself. AR display tech is a cornerstone, but it's just one piece of the equation.
The ML1 might not be the lightfield display we've fantasized about (yeah, me too), but it's the best holographic display you can buy RIGHT NOW. The technology will unquestionably keep improving... but you have to start somewhere.
We didn't magically get to photorealistic 144fps 3840x2560 videogames overnight, and I have a mountain of old hardware in my closets to prove just how far we've come. Enjoy the ML1 for what it is, and look forward to what it's paving the way towards next... and pray that the race to 'cheap' doesn't win out over the quest for 'awesome' before XR technology GETS to awesome.
Part of the reason computing power stalled was due to hitting die-shrinkage limits, but much of it is due to the expectation that modern hardware has to be dirt cheap. I paid around $2500 for an a
Amiga 3000 circa 1992. What kind of a beast of raw, brute-force AMD64 power could you build TODAY with that same inflation-adjusted amount (say, ~$7,000)? If people still routinely bought $2,500-3,200 laptops & tolerated inch-thick 10lb form-factors, what kind of laptop power COULD we have now?
Thin, light, and power-sipping has won for now... and the computer industry needs a reason to say, "fuck all three, give me a backpack-sized mainframe". AR *is* that reason.
The ML1 isn't cheap, but its price really isn't out of line with what you'd spend if you bought an Oculus Rift and a gaming laptop comparable to what's inside the ML1's LightPack controller.
If anything, the single biggest problem with mixed/augmented/virtual reality today is that it really needs way more horsepower than any mainstream (let alone cheap) consumer device currently has. Current hardware is kind of like a NeXT back when it was the computer to die for... lots of promise & future-looking software, running on hardware that just wasn't quite fast enough to satisfy people's expectations.
In all honesty, XR (my favorite umbrella term for mixed/augmented/virtual reality) is what the currently-moribund PC industry NEEDS... an excuse to RADICALLY increase computing power. We haven't had an excuse like that for 10 years. The same beefed-up hardware that will enable realtime XR applications with low latency and fluid animation will finally give us things like "Aero Diamond" (Aero-like Windows graphics, but with realtime-raytraced eyecandy and translucency effects) once even a mid-range laptop has the equivalent of today's most expensive hardware.
NVidia has taken the next step towards realtime hardware-accelerated raytracing, and Intel & AMD have started moving into 8+ core 5+GHz territory. Pair the display hardware of a ML1 or Hololens with a 16-core i9 running at 4.5-5GHz with 64gb of RAM, a 2TB SSD, and a top of the line dual-slot NVidia GPU (call it "personal cloud"), and watch the real magic happen. Pair the same display tech with the equivalent of a high-end Android phone, and prepare to be kind of underwhelmed, just like we were 25 years ago with NeXTSTEP. The fundamental idea is good, it just needs radically more-powerful computer hardware driving it to make it truly awesome.
The only people who think replacing keyboards for general text input is a good idea are people who never bothered to learn how to type.
Mobile devices might be an exception due to size... I'm a Graffiti person myself (originally Palm, now Android). It's how I'm able to make long postings like this one with my phone. Predictive keyboards slow me down, because I'm kind of OCD about not making spelling errors & only Graffiti is accurate enough to let me get away with NOT proofreading everything word-by-word. My only gripe with Android Graffiti has to do with capacitative behavior & governor-induced lag... it pisses me off that a fsck'ing sub-20Mhz 680x0 could handle nearly error-free input, but an Android phone with a faster CPU & more cores occasionally can't tell the difference between 'o' and 'u' because... er... Android decided something BESIDES high-res stroke-capture is a higher priority at that moment (using a governor like 'performance' or 'responsive' on a rooted phone definitely helps).
That said, there's room for improvement with Qwerty. The patents on the Matias Halfkeyboard expired years ago... it should now be a default feature on any new keyboard (and Matias himself should sell a USB adapter to sit inline and turn ANY USB keyboard into one), not to mention a standard feature in Linux and Windows. It's handy for "left hand on keyboard, right hand on mouse" use cases.
At this point, the main point of something like the ML1 isn't just "rendering floating metadata bubbles next to interesting objects", it's "being able to visualize things in 3D, without many of the drawbacks that severely limit the amount of immersive virtual reality someone can take before they start getting queasy."
The fundamental problem with fully-immersive virtual reality is that current VR hardware just isn't fast or responsive enough to sail over the uncanny valley. If you're wearing a ML1 and rendered objects quiver a bit as you look around, your brain interprets it as "hey, there's motion over there... you should probably pay attention to it, just in case it's a tiger hiding behind a log". If you're wearing an Oculus Rift and the entire virtual room around you sloshes, lags, and jerks around, your brain just goes into meltdown because observed reality is now in frank contradiction of everything your OTHER senses are telling it. And if you're wearing a Rift, looking at some 3D virtual object that follows your head motions, but otherwise remains fixed & motionless regardless of YOUR OWN motion, you'll end up with good old-fashioned motion sickness.
Put another way, mixed reality buys you a metaphorical hall pass & loophole to get around your other senses, by ensuring that MOST of what you see is in agreement with your other senses, regardless of whether specific sub-details accounting for a small percentage of your surroundings are in agreement as well. The more your larger environment appears to be consistent with your other senses, the less likely you are to be affected by vertigo.
Go to http://creator.magicleap.com/
You have to sign up for access, and the number of hoops you have to jump through to get into the site for the first time and see anything at all is kind of annoying... but once you're in, the site is quite good, and generally fluff/bullshit-free. It's almost like the development team hid it to keep it safe from the marketing team's interference. It's a total night/day difference from the "main" site... lots of real, meaty content. Anyone can register, and it's definitely worth the effort if you're even slightly curious.
Big tip: make a special effort to recursively work your way down the "Bootcamp in a box" section... it links to other important sections that aren't necessarily obvious or easy to find from the menu on the left (that was 1-2 months ago... it might have improved since then).
When running the various "Hello Cube" variants for the first time, remember that the real device has a limited field of view where holographic content can go, and the simulator faithfully emulates this behavior. If you don't see the cube, look around the whole virtual room VERY carefully and systematically, sweeping through a 360-degree arc approximately between the floor and ceiling. Chances are, the cube IS there somewhere... it might even be located within the part of the room you can see, but outside the zone where holographic content can be.
You also might be too close to the cube to see it -- anything surface closer than ~18 inches gets culled by Unity, and remember, the INSIDE/REAR surfaces of Unity/OpenGL triangles are normally culled & invisible if you're looking at them as well, so if your virtual nose is metaphorically poking into the cube, you probably won't see ANYTHING at all. If you don't see anything after exhaustively & systematically looking around the room in a 360-degree circle, try stepping back ~4 feet and look around the room again.
Pay special attention to the "Planes" and "Raycast" examples. If your biggest question after a couple of days is, "How do I make sure {something} ends up on the table, or sofa, or a specific area of the floor", it's probably because you haven't worked through the Planes & Raycast examples yet, and as a result you're missing a fairly fundamental point of mixed-reality development. With early augmented-reality development for devices like Google Glass, you could skirt around the problem because conceptually, everything you rendered was being displayed on a ghostly planar surface a few feet away. Magic Leap takes it up a few notches, and allows you to choose the apparent distance of an object, too. With freedom and power comes responsibility... being able to place objects semi-arbitrarily within a room means that at some level, your program has to be at least vaguely aware of what's IN the room surrounding you, and take that into consideration when deciding where to put things. You CAN'T just blindly position objects at arbitrary coordinates & expect everything to "just work"... you have to make sure that you're either putting things where nothing already exists in the real world, or be ready to deal with the consequences OF making ghostly holograms coexist within the same volume of space as a real-world object.
Also, keep in mind that the ML1 semi-passively builds its internal model of a room in the background as you look around. This also applies to the simulator. If something involving planes/raycasting to find surfaces doesn't seem to be finding a surface that you KNOW is nearby, help it out... slowly look up and down around the room, and do it from several vantage points. The more data you give it (by looking around the room in a systematic manner, from different angles), the better its model of the room will be.
Update: I was wrong about maritime law & salvage rights. Apparently, the US isn't a signatory to the UN treaties governing maritime salvage, and takes the position that a) sunken US government ships that sink on the high seas remain the exclusive property of the US in perpetuity and are never implicitly abandoned; b) salvage rights to ships that sink in US waters remains with the original owners in perpetuity... and solely belongs to the US government if the wreck gets salvaged by anyone else; and c) jurisdiction over manmade objects in space remains exclusively with the nation-state associated with its launch in perpetuity.
So even IF Hubble were in uncontrolled freefall, if SpaceX fixed Hubble on its own, NASA would (at best) say, "Thanks for fixing Hubble at your expsnse... it's still ours." And SpaceX & its employees *could* conceivably be prosecuted on federal charges for touching Hubble without permission if the feds were determined to show the world that no good deed ever goes unpunished.
TL/DR: The US explicitly rejects most modern international maritime law, and in any case asserts that maritime law does NOT extend to space.
A Trackpoint isn't a touchpad... it's a tiny isometric pointer stick commonly associated with Thinkpads. And the whole point is that it enables you to use the mouse without having to lift your hand from the keyboard (something that the finger-pointing unwashed masses just don't seem to grasp the importance of).
I hate finger-pointers. They're the users who ruined touchpads. Back in "the old days", touchpads acted like flat trackballs, but adjusted curving gestures to appear straight like a thumb-trackball did. Then, manufacturers all jumped on the "optimize touchpads for finger-pointers instead of thumb-sweepers", and touchpads became unusable by thumb-sweepers. They COULD have made touchpads dual-mode, but instead they just eliminated the thumb-optimizing DSP logic entirely. Fuckers, that's why we can't have nice things...
I personally don't get the obsession with eliminating the top, and especially the bottom, bezels. Width affects the max screen size that can fit in a back pocket, but portrait-orientation height? Pffft.
Persondlly, I *like* having a reasonable bottom bezel, because it's really hard to hold a large, heavy phone with one hand & operate the back/home/whatever buttons with my thumb if they're all grazing the lower edge. IMHO, the most grippable phone I ever owned was my old Galaxy S3 with Seidio extended battery case... it put the center of mass comfortably low & provided a handy ridge to grip the phone with from behind. The phone couldn't lie flat, but it had a kickstand which mitigated 99% of the annoyance & overall, it rocked. Even though I eventually bought bigger batteries, I kept going back to the Seidio battery/case because it was SO much easier & more comfortable to hold.
The big question, though... if you keep the phone in an Otterbox Defender-like case & the rear glass cracks, do you HAVE to fix it, or can you just say "fuck it, it's hidden inside the case anyway" & just leave it cracked without affecting anything besides "out of case" aesthetics?
Also, does the glass help or hurt with heat-removal? Can the phone run at 100% max from power-on until battery-drained (say, if you're using Daydream VR), or will it go into thermal-throttling within a few minutes if you try (like a Nexus 6P does)?
We might not eat 100% of a cow, pig, etc... but the net waste is practically zero. AFAIK, practically every cell of a cow now has commercial value for something. A friend who once had a summer job at a slaughterhouse told me that most of the trash leaving a slaughterhouse comes from garbage cans in the offices & break room, and most of the remainder comes from the janitorial or maintenance departments (detergent containers, old knives, etc) & ends up getting recycled.
If you want to try out a fun product that will whet your appetite for a Magic Leap ML1 or Hololens, check out Aryzon's headset (http://aryzon.com). It's basically the augmented-reality evolution of Google Cardboard -- you mount your phone in it, the display's light bounces from a mirror onto reflective window tinting, and you see a 3D hologram superimposed on whatever is in the room in front of you.
Aryzon's lead (only?) developer has done a fantastic job of writing documentation, creating demo apps, and making tutorial videos ( https://www.youtube.com/channe... ).
Make no mistake... it's nowhere close to being in the same league as a ML1 or a Hololens... but at 30 Euros (~$36) with free shipping worldwide (including to the US), it's actually cheap enough to buy for shits & giggles. Don't be afraid to order one just because it involves mailing to the US from the Netherlands... NL-US mail is pretty fast (it also might be available from Amazon, if you found this post via Google long after it was a current topic on Slashdot).
There are two similar products I'm aware of to Aryzon's headset -- HoloKit and the Lenovo Lightsaber:
* Holokit's website (holokit.io) has been dead for at least the past few days, and probably longer. You can get it from Amazon with 2-day Prime shipping... but I'm not really sure what you could DO with it once it arrives, because as noted... their website is missing in action. And AFAIK, HoloKit doesn't have a headstrap, so you have to use one or both hands just holding it up. Speaking as someone who's done the whole "hold a Cardboard viewer with a heavy phone on your face using one hand" routine in the past, I can definitely say that it gets old (and tiring) really fast.
* The Lightsaber is probably a step-up design-wise, but (AFAIK) has no public SDK. Lenovo has apparently been promising one since last year. Enough said. Without a SDK, it's just a silly toy.
Aryzon is a small company, but their main (only?) developer has done an amazing job of writing demo software, creating documentation for developers, and posting tutorials on Youtube (https://www.youtube.com/channel/UCy7KJ7TQ3myJ0fSVqFDudeg/videos).
That said... MagicLeap and Hololens are more than just polished hardware. They both have a substantial amount of software backing them up to do things like 3D surface-mapping and immersive binaural sound. Do you need it if your entire goal is to admire a holographic balloon floating a few feet in front of you? No, not really. But you're really, really going to need it if you want to render something onto a random tabletop, floor, wall, or ceiling in a random room somewhere... because THAT'S a really HARD problem to solve. So hard, in fact, that when you first get into mixed/augmented-reality development, you probably won't even realize it IS a problem until you've been at it for a few days... and then, the sheer enormity of it hits you like a boulder.
TLDR: if $2,295 is out of your budget, buy an Aryzon. It's cheap & fun, and it'll give you an affordable taste of why holographic mixed reality is so cool. Just remember that there IS more to "mixed reality" than "being able to see digital holograms superimposed on whatever's in front of you", the same way there's a world of difference between experiencing 6DoF immersive VR with an Oculus Rift, haptic gloves, and full-body motion tracking and experiencing the equivalent of a Viewmaster where you can stand fixed in one spot and enjoy a 360-degree view by moving your head around.
Better analogy: you own a big, huge, old London-style double-decker tourist bus on a small, remote island. Say, somewhere in the South Pacific. Twenty five years ago, the bus was delivered to the island by a unique amphibious cargo ship capable of surviving the high seas, rolling up onto the beach, and driving it off before departing to haul big things to other remote islands.
The bus is now old and decrepit... but amphibious ships like the one used to deliver your bus no longer exist... they had some really bad accidents (partly due to bad management decisions), were declared to be unsafe, and quickly retired with a huge backlog of scheduled deliveries that are now impossible to make. On the bright side, the island now has twice-daily low-cost jet service to the mainland... great if you're a passenger or need to ship a small package, little consolation if what you REALLY need to ship is a big, huge double-decker bus that won't FIT inside a modern jet.
It's even worse than that. JWST isn't a real REPLACEMENT for Hubble. It'll be able to do things that Hubble can't, but there are even MORE things Hubble can do that JWST won't ever be able to do.
The fact is, we don't have the ability to launch a new satellite as big or as heavy as Hubble... and AFAIK, there are no rockets even under development that will be capable of delivering something as physically BIG as Hubble (or the existing ISS modules) into orbit. Hubble and the existing ISS modules are all we have, and all we're LIKELY to have for DECADES. They're literally irreplaceable within the span of our lives, and as such, deorbiting them is, IMHO, wantonly reckless and irresponsible. If Hubble's telescope functionality dies before we have the ability to send a refurbishment mission, we should be ready to DO the deorbiting mission on 18 months' notice... but always and only as a last-ditch "plan B" if we don't get the ability to launch a robotic or manned servicing & refurbishment mission first.
From what I understand, if Hubble failed completely and became totally uncontrolled tomorrow, it would be at least a decade before it fell far enough to present imminent risk of uncontrolled reentry. If we made even a token attempt to send a robotic mission to boost it into a higher orbit, we could easily add another decade to that. SpaceX might not be ready to fly a refurbishment mission to Hubble within the next 3-5 years... but it probably COULD be ready to do it 6-10 years from now (if it had a firm commitment from NASA), and will probably be capable of doing it within 10-15 years regardless of what NASA does (and knowing Musk, would probably invoke maritime salvage law & refurbish Hubble ITSELF as a commercial venture if NASA couldn't/wouldn't do it).
Frankly, I think 90% of NASA's sense of deorbiting-urgency is precisely BECAUSE they'd rather see things like the ISS and Hubble get intentionally destroyed than risk allowing someone else to metaphorically grab them from the curb before the garbage truck arrives.
I just want to see the FCC adopt more flexible rules governing spread spectrum, esp. with regard to HF and low-bitrate QRP digital modes.
I think it was Bruce who argued a few years ago that EIRP per hz per time-interval (within a larger umbrella applying to instantaneous power & bandwidth) matters more than the name and legal definition of any particular mode.
If anything, the test for Amateur Extra should include things relevant to advanced digital modes, like OFDM, CDMA, FFT calculation, SDR, matrix math, etc.
Maybe set aside a 100MHz sandbox for the lowest license-class hams to play with without restriction (besides input power and total bandwidth) at low power in the 10GHz band (almost nothing you can do with a watt at 10GHz is going to matter much, even if you screw up in nearly every conceivable way), expanding bands & power as you move up the license classes (reserving HF wideband spread-spectrum for amateur-extra class (which is something that WOULD motivate me to move up from General class).
The idea being that someone who makes it to AE will hopefully at least understand HOW things can innocently go wrong with exotic digital modes, so the FCC can give them more freedom to play in situations where allowing a (novice-)Technician class licensee would be like handing a loaded gun to a toddler.
And how, exactly, would compromising a 2m/70cm radio with a range of a few miles likely to be of any use to the Chinese government? It transmits FM audio. ANYONE with an antenna & receiver nearby can receive it, by intent of the radio's owner.
And even IF it could somehow be hijacked by Chinese agents... the last time I checked, China's government has no legal jurisdiction over me (an American citizen in America), and I'm not going to flatter myself into thinking that anything I might do is of the SLIGHTEST interest to them.
There's awareness, there's paranoia, and then there's silliness. If you're going to lose sleep over a $29 pocket radio from China, I sincerely hope you don't own an iPhone, most Android devices, ~97% of the LCD TVs in existence, or anything that combines networking and a microphone. Worrying about the security implications of using a Baofeng radio is like losing sleep about sea-level rise because it rained yesterday.
In the US, from the perspective of common law, the buyer would technically have no recourse in a purely civil claim unless he could establish that he suffered harm that can be quantified in dollars... and that he actually DID suffer a net loss. So, he might still have an actionable claim if he suffered indirect harm whose cash value exceeded the instant profit from the art doubling in value... but it would be an uphill battle to quantify the cash value of an emotion like "disappointment".
There's also & independently the possibility of criminal prosecution ("conversion", for example, might be a tort... but in most jurisdictions, it can ALSO be a criminal offense under certain conditions). The catch with the criminal angle is that the buyer purchased "experimental art" from someone KNOWN for doing things like this, so Banksy could semi-legitimately argue that the unanticipated public shredding was itself integral the art & what was purchased.
TL/DR: legally, it's the kind of case only the lawyers would collectively win. If the buyer is unhappy, their best recourse would honestly be to just flip the painting to someone else & walk away. At most, a vengeful buyer might spend a fortune, waste time testifying in court, give Banksy a metaphorical black eye... and be blacklisted as a potential art buyer going forward. "Experimental art" is, by definition, kind of fucked up by "normal person standards", and that applies to BOTH artists AND patrons.
Actually, I just remembered... dealing with symlinks in Windows is still kind of a mess. A few years ago, I tried writing a Java program that would write a file, then create (or replace) a symlink in another directory so it always pointed at the latest file.
It turned out to be hopeless. Not because Java can't create or modify symlinks (it can), but because Windows has bizarre, fucked-up rules governing who is (and more importantly, isn't) allowed to create & modify symlinks. From what I remember, one of the rules is that normal users can create and modify symlinks (in directories they own), but NOT users who are ALSO members of an 'admin' group, in which case theyre only allowed to do it if the program is running with elevated permissions. For the life of me, I fail to understand why it's ok for a user who's one step above "guest user" to create & modify symlinks in any directory he owns, but taboo for a user who's ALSO an admin -- but not running AS an admin -- to do the exact same goddamn thing in his own "Documents" folder.
I could see Windows enforcing a rule like that if you want to create or modify symlinks in a system directory (or even a directory like "C:\Program Files"), but extending the prohibition to ANY directory SYSTEMWIDE (including the user's OWN documents folder) is just plain fucking stupid.
Windows is littered with stupid rules like that. I remember a few years ago when a (seemingly) minor Windows update instantly broke every C# Microsoft Office extension we had, because Explorer and NTFS both enforce different (and not necessarily consistent) rules, and Office enforces yet another set of rules that's consistent with neither Explorer nor NTFS, and partially depends upon the syntax you use to specify paths. From what I recall, we were trying to read and write to files in a subdirectory of a network share that was mapped to a drive letter... previously, it was OK, but under the new rule, we had to use extended-UNC notation. Only in Microsoft's wacky bizarro world is it a permissions error to try and read "z:\path\to\taboo.dat", but totally OK to do the same thing to "\\?\UNC\somehost\foo\path\to\taboo.dat" when both refer to the EXACT SAME GODDAMN FILE.
The point everyone seems to miss amidst the moral panic is that the risk from consuming an artificially-sweetened 2L bottle of soda might not be ZERO, but it's still a huge net improvement over the known, documented harm likely to arise from the daily consumption of a 2L bottle of soda sweetened with sugar or HFCS.
Daily consumption of 2L of diet soda: theoretical changes in gut bacteria.
Daily consumption of 2L of regular soda: significantly elevated risk of diabetes and obesity.
The worst thing that could possibly happen is if someone who's already slightly obese, who drinks 2L/day of diet soda, were to read this and say "fuck it, it's healthier to drink regular soda instead." Because it's not.
I think it was partly pollution, but also partly the realization that if she hadn't, 99% of the trees within 50 miles of London would have been gone within a century. Wood just doesn't have the energy-density necessary to provide winter heat to big cities with hundreds of thousands of residents... not only does it require lots of time and land to grow, but the logistics of delivering it (in sufficient quantities, with sufficient frequency) become insurmountable as well.
I'm pretty sure that's at least part of the reason why in 1400, cities like Rome, Athens, and Constantinople/Istanbul had been large, sprawling cities for centuries, but cities like London & Paris were still overgrown forts & trading posts. At least in cities like Rome, heating a dwelling was more of a luxury and matter of personal comfort than a matter of literal life and death. It's one thing to be unable to heat your apartment when it's 40 degrees outside at night for a few days per year... it's another matter ENTIRELY when it's 20 BELOW ZERO outside at night, and below freezing during the day, for weeks at a time (eg, London). Cities like London didn't have the luxury of being able to treat heating like... well, a luxury. So it became a factor limiting the city's ability to grow.
Except that, unless you
1. Bend over backwards to configure Windows to ask your "permission" before installing updates,
2. Aren't unlucky enough to have Windows throw up a sudden, "Windows wants to install updates... [REBOOT NOW!] learn more" (with "(learn more)" neither appearing to be an obviously-clickable button nor underlined link, printed in a tiny font, and probably inserted into the middle of a longer sentence that itself is neither clickable nor calls obvious attention to itself... and REBOOT NOW! being the default choice that gets selected if you aren't looking at the screen,didn't notice that Windows has rudely grabbed input focus, and press the 'enter' key).
3. Have been making continuous backups (killing much of the performance benefit to having a fast SSD, unless you ALSO have a fast SSD to use for backups as well)
4. Windows doesn't decide to fuck with your backup drive, too (which has to be continuously connected if you want to satisfy requirement #3).
5. You're ALSO doing backups to an external drive that gets physically removed from the computer and locked away, so it won't be affected by ransomware, or get stolen/destroyed as well if you get robbed or your house burns down).
My biggest beef with pretty much every current Windows backup "solution" is the fact that NONE of the ones that are remotely affordable can gracefully deal with what I call, the size/importance gradient & properly juggle multiple strategies... say, doing frequent incremental backups to a connected drive when the computer is genuinely inactive (say, screensaver active), replicating those local backups to a networked drive on the local LAN when the internal-to-internal backups are finished & the computer is still inactive, and periodically backing up a subset of files from the networked lan backup to "the cloud" (the ones that are important enough to pay ongoing fees to safeguard... I might have ~10-12 terabytes of files, but only a few hundred gigabytes of them are really what I'd be devastated to lose, and only a few gigabytes of them are truly what I'd call irreplaceable... and at least a quarter of THOSE are sitting on remote git repositories somewhere, anyway). Put another way, indiscriminately and continuously keeping 10+ terabytes of files backed up in "the cloud" just isn't sane (in terms of cost OR ongoing performance), but keeping a subset of them appropriately backed up is a major pain.
What I'd LOVE to see from Microsoft (and what would convince me to switch to Linux once and for all, if it appeared there first) -- a new kind of filesystem I'll call "DHFS" ("data-hoarders' file system"). In day to day use, it would work like ext4 or NTFS... files get written directly to the drive, for performance. However, in the background (as a lowest-priority system task), it goes through periodically, finds all the "real" files, moves the "real" file to some "master" volume (assuming there isn't already an authoritative, identical copy of that exact file already there), and replaces it with a DHFSymlink (which works like a "regular" symlink for reading purposes, but writes/updates result in the symlink getting blown away & replaced by the new data (or if it's an append, some special structure that says, "the start of this file is (some file pointed to by this link), but the subsequent data is (this)", since creating an entire new copy of the file just to add a byte to it would totally kill performance).
With something like DHFS, only the authoritative master copies of the files, and their MUCH smaller symlink data, gets backed up.
In effect, DHFS would systematically preserve directly-usable copies of every unique file on the system (assuming you could figure out the "real" name and location of any given file), while preserving the CONTEXT & Metadata of those files with something similar to symlinks. If the ability to roll back state wasn't required, it would also do something akin to mark & sweep garbage collection... it would periodically go through its ar
One big problem I see... as far as I can tell, the standard does nothing to clarify whether a given device supports ONLY 2.4GHz, or whether it supports 2.4GHz *and* 5GHz... and if it supports 5GHz, which channels it supports & how.
There are lots of 802.11ac devices, for example, that either don't fully support the use of U-NII(2C) channels, or have crippled DFS implementations that use a sledgehammer instead of a scalpel... satisfying the FCC's requirements, but doing it in a way that results in a product so crippled it almost might as well not even bother with U-NII(2C) channels(*).
I'm also curious to know how they intend to deal with things like AP-roaming and dynamic handoffs... something that was theoretically defined on paper way back in 2008, but (AFAIK) has NEVER really worked properly with consumer devices on home networks. Or pretty much ANYTHING besides a tightly-controlled Enterprise network.
This is my major beef with wireless network gear today... it's DAMN NEAR IMPOSSIBLE, even if you know EXACTLY what standards you need compliance with, to actually walk into a store like Best Buy and make an informed purchasing decision based on their advertising literature and packaging. And if you DO go online and read teardown reviews, there's still no guarantee... the manufacturer could have completely changed not only the general design, but literally changed out the entire chipset with a completely different one that has inferior performance or standards support while keeping the model number (often, even the UPC) unchanged. Linksys & Netgear are both notorious for this... often, they'll indicate the revision on a sticker on the device itself, but put NOTHING on the packaging that's visible before you break the shrinkwrap to indicate whether you're getting the one that earned 5-star reviews & had people drooling, or the later version that got 1 & 2-star reviews and is a pale, cruel imitation of its earlier self.
----
(*) Many of the 5GHz channels share spectrum with weather radar & have to make a "robust" attempt to detect its presence and refrain from using frequencies where it's detected. The variability comes from the fact that some devices take the cheap approach... shutting down and going dark entirely for at least a minute to listen for radar, and assuming the worst at the slightest hint of a signal. This makes the FCC happy, but results in a product that's dysfunctional (to put it nicely).
The cheap/usual way is for the AP to just "go dark" for a minute while listening on the channel for things like weather radar transmissions. When this happens, the wifi connection appears to just silently drop for no apparent reason, then reappears about a minute later (assuming it didn't detect what it thought was a radar signature).
The more sophisticated way is for an internet-connected device to include a GPS receiver & query the FCC's database directly after discerning its location... if it's not within range of a known radar site, it can skip the majority of physical DFS checks requiring radio silence (basically, doing it once at startup). Alternatively, the AP could include an additional radio receiver & logic so that prior to "going dark" on a DFS channel to listen, it could temporarily switch to an alternate, non-restricted 5GHz channel (and notify clients it's about to change channels). The alternate channel might suck, but IMHO, "works poorly for a minute " is STILL a huge improvement over "goes dark and doesn't work AT ALL for a minute every hour"
I might be wrong, but I think 802.11ad MIGHT (in theory) have the necessary hardware to simultaneously use two non-contiguous 5GHz channels (say, 36 and 104), in which case it could (conceivably) "go dark" on channel 104 for DFS while maintaining an active link with connected clients (at half-throughput) on channel 36... but whether any 802.11ad device you can actually go out and purchase TODAY as a non-Enterprise customer (or any random electronic device conceivably purchased at Best Buy or Walmart) can actually SUPPORT that is anybody's guess.
A message warning about an earthquake arriving in 4 seconds still has merit... if nothing else, it increases the likelihood that your phone (and for women, your purse) will be in your hand (rather than 7 feet away) when the earthquake strikes, maximizing your likelihood of being able to use it to call for help if you get partially buried under debris (or at least ensuring that you're ready to grab the phone and/or purse a moment later, instead of having to stop & think about it).
> The GSM protocol includes a broadcasting feature that overrides all other transmission
> in order to deliver emergency messages to all cellphones simultaneously.
> Why then the delays?
Because roughly half the cell phones in America have historically been CDMA devices, not GSM. CDMA generally had comparable functionality (on paper, at least), but wasn't literally identical.
Compounding the problem, major parts of CDMA's functionality was officially "optional" & left up to the carrier to pick & choose. Qualcomm intentionally allowed Sprint & Verizon to implement CDMA in slightly-incompatible ways... and Sprint & Verizon liked that, because it meant that even a theoretically-unlocked phone from one network would be forever crippled & dysfunctional on the other, EVEN IF a user managed to get it activated somehow.
Ultimately, it was (mostly) Apple & Google who put an end to much of the silliness. Blackberry & Sidekick mitigated it... but only for THEIR devices... and used their mitigations as a way to try and lock out Palm & Microsoft. The main thing that saved Apple & Google was Microsoft's purchase of Danger & subsequent willingness to license out their patent pool on fair & non-discriminatory terms (and why Microsoft makes more in profits from the sale of an iPhone or Android phone than it ever did from the sale of a Windows Mobile phone).