Maybe, but that 0.001% has tremendous influence on the purchasing decisions of OTHERS. Who do people looking for a new phone ask for advice? The 0.001% who root & reflash their phones. If THOSE users think Huawei is now the devil & tell everyone who asks for advice that Huawei phones suck, they're unlikely to buy Huawei phones.
Think back to how quickly Nokia went from #1 worldwide to "basically irrelevant". What happened? They made a business decision to ignore the US market since American GSM was a hot mess (T-mobile barely had enough spectrum to do 2G GSM with 19.2kbps GPRS data in most US markets, most of AT&T's high(er)-speed data was EDGE, not 3G, and AT&T's 3G was 850MHz, not 1900/2150MHz), and they didn't sell many phones in the US anyway compared to even small countries like Portugal & Ireland.
What Nokia overlooked was mindshare & influence. America might have been a minor market, but it was a hugely INFLUENTIAL market. When Nokia phones disappeared from American stores, Nokia phones ALSO disappeared from the blogsphere, magazines, and review sites (the majority of which at the time were, in fact, American). The fact that the few Nokia phones that occasionally showed up as expensive imports in places like New York & Miami (intended mostly for foreign visitors to buy while on vacation & take home) were almost USELESS in the US just made matters worse, and got them written off as 'irrelevant' by even more American tech writers. (True story: ~10 years ago, Nokia had a store in Miami at Dadeland Mall whose primary market was visitors from Latin America. Every single one of their employees had non-Nokia phones for personal use, because they literally didn't have a single Nokia phone in the entire store that was capable of EDGE on ANY network, or 3G on AT&T. They were "1900/2150 3G + GPRS, take it or leave it" (and everyone who lived in Miami opted for "leave it").
How big was your TV? As a practical matter, for 3D Hollywood movies to give you the full effect intended for theater viewing, the screen HAS to fill your entire FoV, because near objects are larger than distant objects on the screen, and the nearest an object can GET is the distance at which the object would fill the screen from edge to edge. So yeah, a 52" TV sitting 10 feet away on a high stand is going to suck for 3D. With a 65" or larger TV, viewed from 5-6' away in a darkened room, the effect is MUCH more impressive.
This is why it's so important for people buying cutting edge technology to UNDERSTAND its limits & constraints. If you're just running into them randomly while the vendor pretends they don't exist, you're probably going to have a disappointing experience with it. If you understand exactly what those constraints are, how to mitigate them in actual use, and go into it with realistic expectations, you can enjoy it. That's also why first-gen hardware is always targeted towards developers... developers will actually read the 400-page instruction manual & have at least some understanding of the technology's capabilities & constraints. Average consumers won't even read a 4-page quick-start guide (though in their defense, that's usually because they tend to be almost completely devoid of useful information... if someone looks at the instructions and the first 2 pages are legal disclaimers, copyright notices, and general bullshit nobody actually cares about, they aren't likely to keep searching for the paragraph or two of actual information carefully hidden somewhere among it.
I don't disagree about the inadequacy of 90hz, I was mainly pointing out that even with hardware specs nominally comparable to a top-shelf Android phone, Oculus does still have value due to its superior heat-removal vs a phone.
The main problem with 3D television is that the TV industry had already given up on it and walked away by the time Blu-Ray finally got its shit together and had actual hardware consumers could go out and buy.
It's not hopeless, though. Eventually, we WILL have TVs with the equivalent of FreeSync that can be driven directly at 96-144hz over HDMI. Once that happens, it'll just be a matter of time until manufacturers of Blu-Ray players add the ability to control 3D shutter glasses into the player itself. Explicit support for "3D" from the TV itself won't be necessary, because all the required logic can be moved into the player as long as the TV can be driven directly at 96-144hz.
Dev kits really need to be over-engineered and over-spec'ed compared to consumer-oriented hardware, especially early in a product family's life. A company that releases a brand new product, then renders it completely obsolete within 12 months will piss off a LOT of people, so you need to ensure that the first-gen hardware has at LEAST enough horsepower to limp along with the likely next-generation hardware. People grumble when the expensive hardware they buy today sucks at running software written for next year's hardware... but get downright PISSED and ENRAGED when the expensive hardware they buy today won't run software written for next year's hardware AT ALL. Premature value-engineering is the kiss of death for new platforms.
The main thing Oculus has going for it compared to, say, a high-end Android phone in a head bracket, is heat-removal. Even phones that technically have the horsepower to do reasonable VR just can't "take the heat". Case in point: a Nexus 6p. On paper, it has impressive specs... but if you ignore Google's cautions and try running Daydream on it, you'll hit the 6p's heat limit within minutes, thermal-management throttling kicks in, and the whole use experience rapidly goes down the toilet. An Oculus Rift isn't expected to be pocketable or have long battery life, so even if it HAS nearly the same specs as a decent Android phone, the difference is that it can run a marathon using that hardware instead of going into metaphorical heatstroke within a few minutes.
AFAIK, the VFX-1 was REALLY the equivalent of wearing 3D glasses in a darkened movie theater with a perfectly-positioned seat, and Comanche didn't do head tracking at all. 30fps for any application where you're expected to be able to turn your head & have the screen immersively track the motion in realtime would be completely INTOLERABLE after ~5 minutes. Hell, I could barely endure the gen1 Oculus Rift demo (the one where you're seated in front of a virtual desk looking around) without getting queasy & a headache.
I maintain... for any application where, at the minimum, users are expected to be able to immersively experience the equivalent of sitting in a chair and looking around a room through a welder's mask (with everything the user sees being synthesized and displayed on video screens in front of the eyes), 90-100fps is the absolute rock-bottom minimum for tolerability... and really is nowhere close to being fast enough to be satisfying once the novelty wears off.
Immersive first-person shooters make for cool prototypes and demos, but IMHO, the first mixed-reality games that will be genuinely fun and playable will be modern ports of games like Battle Chess, Archon, and Sim City (where there's zero immersion and a very fixed, stable point of reference, and no question that you're viewing a computer-generated hologram rendered onto a tabletop).
We all still have a lot to learn about what makes virtual/augmented/mixed-reality experiences enjoyable, what makes them unpleasant, and how to mitigate things from the 'unpleasant' side within the constraints of practical hardware. Compare a FPS from the early 2000s to one from today. I remember a period of a few years circa 2005 when most FPS-type games had become almost unplayable (without major vertigo) by anyone over the age of 12, because they'd become photorealistic enough to SEEM like they should be immersive, but REALLY did SO MANY THINGS "wrong" (from the perspective of sensory overload, logical contradiction, and ignorance of basic physics), it's almost a miracle the genre even SURVIVED (younger kids weren't as bothered, because they tend to just take things at face value instead of constantly scrutinizing the nature of reality around them).
Eventually, certain design patterns emerged that allowed games to subtly anchor and ground themselves into the player's surroundings and reality, so your brain could once again look at the screen & say, "ah, this is cool-looking... but not real, so there's no need to freak out about sensory contradictions". For at least the next few years, the key to making successful games that take advantage of virtual, augmented, and mixed-reality will be resisting the temptation to jump straight into "you're immersed in an imaginary world", and instead settling for "virtually enhancing the world already around you".
With high latency, you can't even stand in place & look around a fully-immersive VR environment without vertigo becoming a problem. So no, higher framerates aren't a cure-all, but at LEAST they enable nausea-free "look around while standing still" immersive VR.
Simply put, your brain can deal with small objects in a large scene that have wonky motion, but draws the line when seemingly EVERYTHING, including your larger surroundings, is in sensory conflict.
VR needs MUCH higher framerates & close to zero latency for any application that involves immersive "look around by turning your head" navigation. 60hz just doesn't cut it. Even if you read your sensors & render frame {n+1} while displaying frame {n}, you're looking at ~17ms of latency *anyway*. Anything longer than 10ms feels sloppy (like sloshing around in water), and anything longer than 5ms is perceptible. At 90-100hz (used by most official VR platforms), you're on the bare outer edge of tolerability.
Likewise, you need ultra high resolution... at least, in the specific spots you're looking at. In theory you could use eye-tracking, but the industry is still working on ways to focus detail where it's needed without making the scene visibly shimmer in peripheral vision (the eye is simultaneously picky & oblivious to higher-order artifacts in ways we're still discovering.)
To move to the next level, immersive VR needs:
* faster framerates. As in, 200hz or better.
* GPU hardware that can render a frame at double the video framerate. No "mobile" GPU comes anywhere close today, and a 2-slot desktop card barely approaches it.
* short-range wireless networking with sub-millisecond latency & gigabytes per second of bandwidth (so you can do your rendering with a beefy mini-supercomputer across the room).
In many ways, MagicLeap's technology sidesteps some of these problems. For example, instead of having to completely render an entire scene from scratch, it can focus on just the holographic part, then use straightforward video techniques to optically anchor it to an object in the room (making the framerate of the hologram less important to the immersiveness experience). In other words, with ML, you can look back & forth, and the scene around you doesn't 'slosh' because you're seeing the room itself directly, and the hologram just appears to wobble slightly. Since the BIG detail (the room) doesn't lag, your brain is more tolerant of the hologram's quirks.
Virtual/Augmented/Mixed Reality is now approaching its "Roomba" stage... finding ways to do useful things with presently-viable technology. We're a LONG way from general-purpose robotic servants (or a Star Trek Holodeck) & still have to hand-scrub behind the toilet, but at least the cat's daily furballs & tracked litter get cleaned with a button-press (and the equivalent of child-proofing the room to accommodate the robot's limits).
As far as the overwhelming majority of Americans are concerned, cars ARE king, and pedestrians are the intruders. You can debate whether that's good or bad, but the fact is, even people who LIVE in the urban cores of big cities almost always own cars unless they're prevented from doing so by virtue of poverty (or they use cabs and Uber SO OFTEN, the distinction between owning and not-owning a car is almost a matter of abstract semantics).
From the standpoint of that overwhelming majority, transit has actual MERIT only if it promises (truthfully or not) to reduce traffic congestion, and any form of transit that's likely (or practically GUARANTEED) to make traffic congestion worse is politically intolerable.
The utter failure of transit planners to come to terms with this reality is why Miami voters feel like victims of an elaborate bait & switch scheme every fucking time we get tricked into voting to increase transit funding. The county transit agency runs a multi-month advertising and public relations campaign that makes it look like they're planning to extend Metrorail down to Homestead, up to Broward, to South Beach, and out to West Dade... then six months later, quietly spends all the money on buses.
Fact: if Miami-Dade Transit had been 100% honest about how they intended to spend new transit funds (eg, on buses), voters would NEVER have agreed to either of the past two transit tax referendums. They would have gone down in flames and been defeated SO BADLY on election day, half the transit department's planning staff would have probably been laid off the following day in retribution.
Making matters worse, Miami's transit agency has an absolute KNACK for taking a good idea, then turning it into a proposal that's so fundamentally-flawed and awful, even people who SUPPORT building new rapid transit lines can't stomach it.
Specific examples I remember from recent years:
* The "East-West" Metrorail line, proposed to extend Metrorail from its current orange-line terminus at Miami International Airport west along State Road 836 (more or less) to Dolphin Mall, the south to Florida International University. Slam-dunk, right? Wrong. At one point (around 2007), they drew up fairly specific plans that showed where they planned to build stations. Let's just say that they literally couldn't have come up with worse proposed locations if they'd explicitly TRIED to achieve that as their primary design goal. Take the proposed station "at NW 87th Avenue". OK, solid choice... a square mile of medium-density office park to the northeast, a square mile of medium-density apartments and condos to the southeast, another square mile of medium-density apartments & condos to the southwest, and a HUGE lake with no room to build anything meaningful within walking distance to the northwest. Guess where they proposed putting the station? Yep... the northwest side of 836 and 87th Avenue. Literally ANYTHING of pedestrian interest would have been at least a half-mile walk away from the station. You could give Miami-Dade Transit a barrel full of fish & a shotgun, and they'd manage to explosively amputate their own foot. Any HALFWIT could look at Google Maps & realize that the sane place to put the station would be above the CSX railroad tracks, between NW 87th Avenue and NW 82nd Avenue (which would put much of the area to the north and south of the station within direct, easy walking distance, including the residential area south of 836 since 82nd Avenue now connects on both sides (it didn't back in 2007, but the reconstruction of SR836 was ALREADY underway, and if MDTA planners DIDN'T know about it, they sure as fucking hell SHOULD have).
* An EVEN WORSE plan to extend Metrorail west along SW 8th Street instead of 836... a route that would have not only managed to completely miss two huge malls (International Mall & Dolphin Mall) by 2 miles (as opposed to putting stations on the edge of their respective parking lots), but would have put two or three $50-100 million stations within conveni
The US is huge, but the population isn't all evenly-distributed. Look at a night satellite image of the US... the lights will show you where most Americans REALLY live.
Take Florida... vast & sprawling, but ~90% of Floridians live within 10 miles of I-95, I-4, or I-75. Ditto for Georgia, where ~80% of the state lives in Atlanta or Macon.
If "prime" property is used as a surface parking lot, its "prime-ness" is more a case of wishful thinking. Most new skyscrapers (the TRULY prime property) have their parking sandwiched between ground-floor retail & rooftop resort-like park/garden in the 10-20 story pedestal below the tower itself. One of the ironies of downtown Miami is that an average tower built within the past 20 years actually has more parking spaces per leasable square foot than Sawgrass Mills (a huge outlet mall on the Everglades-fringe of Fort Lauderdale). Planners *hate* it because they think pedestal garages are ugly, but they do mostly solve the parking problem without wasting surface/roof area.
This is why Musk's tunnel-boring is so important. Transit often makes gridlock on city streets WORSE... unless it has its own dedicated right of way. Tunnels are the least-objectionable way to do that, but traditional methods are just too expensive to do it large-scale.
With cheap tunneling, you CAN pitch transit to NIMBYs by saying, "it might reduce gridlock, and at LEAST won't make it even worse than it already is". You can't say that honestly about buses, streetcars, etc.
It's a shame the ADA made future transit projects based on overhead suspended cables impossible (no way to provide a 3-foot wheelchair-accessible egress path from vehicles... floor hatches with unrolled ladders to climb down aren't legal anymore), because it eliminated potential cheap solutions that could literally run overhead with widely-spaced support towers (especially as a way to increase the "reach" of subway stations by a mile or two perpendicular to the main line).
Musk's subways can potentially do that. You could take a subway station that was built in a less-than-ideal location (with a major trip-generator a mile away), and build an automated mini-subway a-la-Musk to shuttle people between that destination and the subway.
Illustrative example: downtown Miami. When Metrorail opened 35 years ago, the "downtown" station was a good half-mile west of what most of what Miamians considered "downtown", and the station in Brickell (Miami's original financial district) was ~2/3 mile from all the tall buildings. The elevated PeopleMover ("Metromover") VASTLY increased the reach of both stations & made most of downtown Miami accessible via Metrorail. Without Metrorail, Metromover would have been a useless toy. Without Metromover, 80-90% of the people who voluntarily used Metrorail (despite owning cars) to commute to downtown Miami in the 80s and 90s would have driven instead.
Modern front-loading washers really DO beat up on their passive components a lot harder than old ones. Old washing machines mostly rotated in one direction (or paused between changing directions), and had a fairly limited repertoire of motions. New ones, with digitally-controlled motors capable of acting like huge stepper motors constantly make radical changes to both direction and speed put ENORMOUSLY more stress on things like bearings, as well as the motors themselves. They also generally spin a LOT faster than old ones ever did, which doesn't just stress out the bearings and springs from increased use... it ALSO increases the likelihood that a slightly-unbalanced load will beat up on those components as it tries to accelerate up to its target rotation speed.
Seriously, just WATCH a modern front-load washer go through its motions in a typical load, then go to a laundromat & compare it to the typical motions of a 5-10 year old commercial front-loading washer. It's almost a miracle that home washing machines are capable of lasting for FIVE years without breaking catastrophically, given the forces they're subjected to when operating.
Combine the increased demands put upon their components with a general trend towards cost-reduction, and the breakage rate of modern washers is no great surprise.
There are several things that need to happen in order to get a GPS fix:
1. You need the satellite ephemera data. Android and IOS phones normally download this over the internet, because it's just a chunk of metadata. It's technically available directly from the satellites, but it takes about 10 minutes to receive a complete copy (the satellites only send a few bits at a time, interleaved between their timestamped beacon data). That's why old GPS devices took so long to get their first fix after powerup, and why they didn't take nearly as long if you kept them continuously powered on and someplace where they could get a clear signal.
2. You need a radio that can receive the satellite broadcasts. The silicon needed to do this first appeared around 15 years ago. The satellite broadcasts themselves are little more than timestamped beacons.
3. You need a way to correlate the timestamped beacon data with the ephemera data. This is where AGPS originally came in. To cut costs, the first phones with "GPS" worked by putting the radio receiver on the phone, but offloading the number-crunching to a remote server. Along the way, "AGPS" did other things that expedited the location lookup... if you know your APPROXIMATE location, it's a lot quicker to calculate your GPS-derived location than it is if you literally have no idea where you are. The more precisely you can approximate your location, the faster and easier it will be to calculate the real location. Since towers were at fixed, known locations, AGPS could say, "Hey, here's the radio data, and we know the user is somewhere within a few miles of (lat,lon) by virtue of using {some specific tower}".
Nowadays, what happens is more like this:
1. You reboot your phone, and it uses the internet to grab a fresh copy of the ephemera data.
2. Your phone knows which tower it's using. It does another query over the internet, and finds out the latitude and longitude of the tower. If it sees two or more towers, it can narrow down your location to approximately 1km.
3. Using the data from steps 1 and 2 to expedite the process, it does its own number-crunching on the GPS data.
In theory, a typical Android phone is capable of getting a GPS fix 100% offline... but it doesn't always work due to poorly-implemented software. Some devices don't even TRY to operate completely offline... if they can't access the internet, their software doesn't even TRY to obtain a GPS fix "the old fashioned way". Other devices have buggy firmware... say, they might fetch a copy of the ephemera data at bootup, but might never try to update it any other time... so if you go for a few weeks without rebooting your phone, the ephemera data will eventually become "stale", and your phone's ability to get a GPS fix will get worse and worse until you finally reboot.
Google changed the rules again a couple of years ago when they added location services to Google Play Services for Android. As I understand it, Play Services attempts to use all sources of location data available to it... if the phone is offline, it should (in theory) be able to fall back to raw, offline GPS... but that will ONLY work if the vendor who sold the device and built its firmware actually IMPLEMENTED offline GPS properly. Because it's so hard to test rigorously outside of lab conditions, a lot of custom kernels (and quite a few STOCK kernels) fuck it up and don't quite get it right (or even try at all).
In any case, on a modern Android phone, the "GPS" radio (which, on newer phones, can almost always handle Glonass as well... and probably BeiDou and Galileo, though the latter two constellations aren't quite 100% functional yet, and I have yet to see a stock ROM that even TRIES to make use of either one) is just one of several sources of location data. A lot of people call the way modern Android phones do GPS "AGPS" because of how it fetches the ephemera data and tower locations over the internet, but strictly speaking, it's NOT the same thing that "AG
Newer cell phones sniff wi-fi beacons & use crowdsourced services that correlate wifi ssids & relative signal strength to locations.
Actual GPS works poorly, or not at all, indoors (besides LITERALLY next to a window), and accuracy is poor because what little signal you get has so much multipath interference.
Mostly, because Condorcet logic is really hard to explain to normal people, and even HARDER to pitch to the media as something that can be presented via soundbites and headlines. My hybrid scheme makes the first round a relatively low-key internal affair for the House of Representatives, then gives the media 3 candidates to talk about & handles the Speaker's election in a relatively straightforward (by American standards) manner beyond that.
The main benefit of using Condorcet rules for round 1, and having the entire vote for Speaker via secret ballot, is ensuring that whomever ends up winning as speaker probably WON'T be the preferred first choice of either party... and that any Speaker who pisses off too many Representatives by getting overly heavy-handed about enforcing party discipline won't be re-elected as Speaker.
There's a second reform I can think of that would severely limit the power of party leaders over individual representatives: whenever a bill is defeated, there's an immediate and automatic secret "no confidence" vote among representatives. If the Speaker loses the vote, a new Speaker is elected immediately (under the same rules as above), and the newly-elected Speaker is not bound by the previous Speaker's committee appointments or policies.
This would put a stop to Speakers who ram bills through the house with single-vote victories by putting the Speaker's position in EXTREME peril if he allows a vote to proceed without being REALLY confident of a solid victory. It might work once... or twice... but eventually, s/he's going to piss off one Representative too many, the bill will be defeated, and the Speaker will be defeated as well. Every Speaker would have to choose between loyalty to his/her party, and desire to keep being Speaker. The relative power of parties to dictate legislation would probably ebb and flow, but any party that pushed TOO hard to control its Speaker would find itself having to continually re-establish its power over new Speakers -- each of whom would be harder to control than the last.
One good way to partially fix the House of Representatives would be to change the way the Speaker is elected:
1. Allow Representatives to vote for ANY House member to be Speaker... by secret ballot, using a Condorcet method. No nominations, and members can't refuse if they win (otherwise, parties would just punish any representative from their party who accepted the position without their approval). Representatives can vote for as many or as few members as they like, and indicate different preferences for those they vote for (with everyone they don't vote for at all being treated as "last choice, with equal preference... so even someone who loyally supports his or her party's choice for Speaker would, at a minimum, have to vote for everyone in the party... assigning the party's choice as their #1 choice, everyone else in the party as their #2 choice, and everyone else as choice #3).
2. The top three candidates from step 1 run against each other, once again via secret ballot among House members. If one of them gets a simple majority, he/she's the new Speaker. Otherwise...
3. The top two candidates from step 2 have a run-off election (also by secret ballot). If one of them gets a simple majority, he/she's the new Speaker.
4. If, however, step 2 produces a result where the top candidate wins a plurality & the remaining two are tied, or if step 3 produces a tie, step 2 is repeated... but this time, under Condorcet rules (as per step 1).
Electing the Speaker this way wouldn't be likely to result in a Speaker who's from a party different from a majority of Representatives... but it WOULD effectively throw a monkey wrench into either Party's ability to enforce party discipline on Representatives, and quite probably result in the election of Speakers who are absolutely, positively NOT the first choice of the Party's own leadership. A Speaker who gets to be TOO heavy-handed about bringing representatives in line would be unlikely to win again, because he'd ultimately piss off too many members of his own party. By keeping the votes for Speaker secret, Representatives could freely vote against those who've pissed them off or antagonized them without fear of reprisal or punishment by the Party.
Be glad you don't have astigmatism, or you'd find yourself in the position of being unable to see clearly without glasses (or eventually, without multifocal lenses) at ANY distance, near OR far.
Ever since roughly age 40, I've felt like I need a binocular microscope to solder anything smaller than 100-mil pitch... and depending on the part & lighting, even 100-mil has left me feeling like I'm "soldering blind" half the time. I've gotten to where I need a magnifying glass just to tell the difference between red and orange bands on a resistor, even WITH bright high-CRI lighting.
Because, ya know, SOMEONE needs to be responsible for sitting on the ISS and tattling on American & European crew members who spend Friday night over on the Russian side sharing a newly-arrived bottle of quality vodka. Or maybe even taking it upon him/herself to legally enforce temperance in orbit by confiscating and destroying any contraband vodka that might arrive on a Soyuz.
I partially jest... but the sad thing is, I can almost see something that stupid and petty happening.
The problem isn't that the water is too warm to run the plant. The problem is that heating lake water to 100+ degrees (F) will kill just about everything that lives in the lake (except algae, which will bloom like crazy and suffocate whatever the higher water temperature didn't kill directly), so they throttled back the plant to avoid overheating the lake.
100 watts is NOTHING if you have a gaming laptop or mobile workstation w/desktop CPU. The GPU *alone* can suck down more than 100 watts, and even when sipping power, you'd be forced to choose between "running" ("limping?") and "charging".
It probably ALSO includes a photographic printer (for high-res, high-quality copies), which would increase the complexity and weight quite a bit. Even if it did Kodak-style B&W "activator" development, the print path would have been pretty complex since the goal was "push-button, print-comes-out" simplicity for users.
I remember those readers... even back in the early 90s, my university charged $1.00/page to print microfilm & microfiche, back when normal copiers were 10c-25c/page (sometimes, 5c-9c if you had a high-value copycard). The only bright spot was that you could sometimes fit 2 pages on one print (microfilm) or 2x2 pages (microfiche) since the prints themselves were so high-resolution.
Do browsers NOT cache static content anymore? Or, in copyright frenzy, do sites mark & tag shit like JQuery, webfonts, etc as the browser-equivalent of "copy-never" to explicitly prevent it from being cached (even within the same site, let alone sites linked to the same static content [like JQuery, Bootstrap, etc.])?
It seems like the kind of stuff that *used* to be aggressively-cached (across multiple sites, no less) by browsers like MSIE & Firefox now LITERALLY gets fetched over and over again).
Slashdot Mirror
> Only .001% of customers unlock their handsets.
Maybe, but that 0.001% has tremendous influence on the purchasing decisions of OTHERS. Who do people looking for a new phone ask for advice? The 0.001% who root & reflash their phones. If THOSE users think Huawei is now the devil & tell everyone who asks for advice that Huawei phones suck, they're unlikely to buy Huawei phones.
Think back to how quickly Nokia went from #1 worldwide to "basically irrelevant". What happened? They made a business decision to ignore the US market since American GSM was a hot mess (T-mobile barely had enough spectrum to do 2G GSM with 19.2kbps GPRS data in most US markets, most of AT&T's high(er)-speed data was EDGE, not 3G, and AT&T's 3G was 850MHz, not 1900/2150MHz), and they didn't sell many phones in the US anyway compared to even small countries like Portugal & Ireland.
What Nokia overlooked was mindshare & influence. America might have been a minor market, but it was a hugely INFLUENTIAL market. When Nokia phones disappeared from American stores, Nokia phones ALSO disappeared from the blogsphere, magazines, and review sites (the majority of which at the time were, in fact, American). The fact that the few Nokia phones that occasionally showed up as expensive imports in places like New York & Miami (intended mostly for foreign visitors to buy while on vacation & take home) were almost USELESS in the US just made matters worse, and got them written off as 'irrelevant' by even more American tech writers. (True story: ~10 years ago, Nokia had a store in Miami at Dadeland Mall whose primary market was visitors from Latin America. Every single one of their employees had non-Nokia phones for personal use, because they literally didn't have a single Nokia phone in the entire store that was capable of EDGE on ANY network, or 3G on AT&T. They were "1900/2150 3G + GPRS, take it or leave it" (and everyone who lived in Miami opted for "leave it").
How big was your TV? As a practical matter, for 3D Hollywood movies to give you the full effect intended for theater viewing, the screen HAS to fill your entire FoV, because near objects are larger than distant objects on the screen, and the nearest an object can GET is the distance at which the object would fill the screen from edge to edge. So yeah, a 52" TV sitting 10 feet away on a high stand is going to suck for 3D. With a 65" or larger TV, viewed from 5-6' away in a darkened room, the effect is MUCH more impressive.
This is why it's so important for people buying cutting edge technology to UNDERSTAND its limits & constraints. If you're just running into them randomly while the vendor pretends they don't exist, you're probably going to have a disappointing experience with it. If you understand exactly what those constraints are, how to mitigate them in actual use, and go into it with realistic expectations, you can enjoy it. That's also why first-gen hardware is always targeted towards developers... developers will actually read the 400-page instruction manual & have at least some understanding of the technology's capabilities & constraints. Average consumers won't even read a 4-page quick-start guide (though in their defense, that's usually because they tend to be almost completely devoid of useful information... if someone looks at the instructions and the first 2 pages are legal disclaimers, copyright notices, and general bullshit nobody actually cares about, they aren't likely to keep searching for the paragraph or two of actual information carefully hidden somewhere among it.
I don't disagree about the inadequacy of 90hz, I was mainly pointing out that even with hardware specs nominally comparable to a top-shelf Android phone, Oculus does still have value due to its superior heat-removal vs a phone.
The main problem with 3D television is that the TV industry had already given up on it and walked away by the time Blu-Ray finally got its shit together and had actual hardware consumers could go out and buy.
It's not hopeless, though. Eventually, we WILL have TVs with the equivalent of FreeSync that can be driven directly at 96-144hz over HDMI. Once that happens, it'll just be a matter of time until manufacturers of Blu-Ray players add the ability to control 3D shutter glasses into the player itself. Explicit support for "3D" from the TV itself won't be necessary, because all the required logic can be moved into the player as long as the TV can be driven directly at 96-144hz.
Dev kits really need to be over-engineered and over-spec'ed compared to consumer-oriented hardware, especially early in a product family's life. A company that releases a brand new product, then renders it completely obsolete within 12 months will piss off a LOT of people, so you need to ensure that the first-gen hardware has at LEAST enough horsepower to limp along with the likely next-generation hardware. People grumble when the expensive hardware they buy today sucks at running software written for next year's hardware... but get downright PISSED and ENRAGED when the expensive hardware they buy today won't run software written for next year's hardware AT ALL. Premature value-engineering is the kiss of death for new platforms.
The main thing Oculus has going for it compared to, say, a high-end Android phone in a head bracket, is heat-removal. Even phones that technically have the horsepower to do reasonable VR just can't "take the heat". Case in point: a Nexus 6p. On paper, it has impressive specs... but if you ignore Google's cautions and try running Daydream on it, you'll hit the 6p's heat limit within minutes, thermal-management throttling kicks in, and the whole use experience rapidly goes down the toilet. An Oculus Rift isn't expected to be pocketable or have long battery life, so even if it HAS nearly the same specs as a decent Android phone, the difference is that it can run a marathon using that hardware instead of going into metaphorical heatstroke within a few minutes.
AFAIK, the VFX-1 was REALLY the equivalent of wearing 3D glasses in a darkened movie theater with a perfectly-positioned seat, and Comanche didn't do head tracking at all. 30fps for any application where you're expected to be able to turn your head & have the screen immersively track the motion in realtime would be completely INTOLERABLE after ~5 minutes. Hell, I could barely endure the gen1 Oculus Rift demo (the one where you're seated in front of a virtual desk looking around) without getting queasy & a headache.
I maintain... for any application where, at the minimum, users are expected to be able to immersively experience the equivalent of sitting in a chair and looking around a room through a welder's mask (with everything the user sees being synthesized and displayed on video screens in front of the eyes), 90-100fps is the absolute rock-bottom minimum for tolerability... and really is nowhere close to being fast enough to be satisfying once the novelty wears off.
Immersive first-person shooters make for cool prototypes and demos, but IMHO, the first mixed-reality games that will be genuinely fun and playable will be modern ports of games like Battle Chess, Archon, and Sim City (where there's zero immersion and a very fixed, stable point of reference, and no question that you're viewing a computer-generated hologram rendered onto a tabletop).
We all still have a lot to learn about what makes virtual/augmented/mixed-reality experiences enjoyable, what makes them unpleasant, and how to mitigate things from the 'unpleasant' side within the constraints of practical hardware. Compare a FPS from the early 2000s to one from today. I remember a period of a few years circa 2005 when most FPS-type games had become almost unplayable (without major vertigo) by anyone over the age of 12, because they'd become photorealistic enough to SEEM like they should be immersive, but REALLY did SO MANY THINGS "wrong" (from the perspective of sensory overload, logical contradiction, and ignorance of basic physics), it's almost a miracle the genre even SURVIVED (younger kids weren't as bothered, because they tend to just take things at face value instead of constantly scrutinizing the nature of reality around them).
Eventually, certain design patterns emerged that allowed games to subtly anchor and ground themselves into the player's surroundings and reality, so your brain could once again look at the screen & say, "ah, this is cool-looking... but not real, so there's no need to freak out about sensory contradictions". For at least the next few years, the key to making successful games that take advantage of virtual, augmented, and mixed-reality will be resisting the temptation to jump straight into "you're immersed in an imaginary world", and instead settling for "virtually enhancing the world already around you".
With high latency, you can't even stand in place & look around a fully-immersive VR environment without vertigo becoming a problem. So no, higher framerates aren't a cure-all, but at LEAST they enable nausea-free "look around while standing still" immersive VR.
Simply put, your brain can deal with small objects in a large scene that have wonky motion, but draws the line when seemingly EVERYTHING, including your larger surroundings, is in sensory conflict.
VR needs MUCH higher framerates & close to zero latency for any application that involves immersive "look around by turning your head" navigation. 60hz just doesn't cut it. Even if you read your sensors & render frame {n+1} while displaying frame {n}, you're looking at ~17ms of latency *anyway*. Anything longer than 10ms feels sloppy (like sloshing around in water), and anything longer than 5ms is perceptible. At 90-100hz (used by most official VR platforms), you're on the bare outer edge of tolerability.
Likewise, you need ultra high resolution... at least, in the specific spots you're looking at. In theory you could use eye-tracking, but the industry is still working on ways to focus detail where it's needed without making the scene visibly shimmer in peripheral vision (the eye is simultaneously picky & oblivious to higher-order artifacts in ways we're still discovering.)
To move to the next level, immersive VR needs:
* faster framerates. As in, 200hz or better.
* GPU hardware that can render a frame at double the video framerate. No "mobile" GPU comes anywhere close today, and a 2-slot desktop card barely approaches it.
* short-range wireless networking with sub-millisecond latency & gigabytes per second of bandwidth (so you can do your rendering with a beefy mini-supercomputer across the room).
In many ways, MagicLeap's technology sidesteps some of these problems. For example, instead of having to completely render an entire scene from scratch, it can focus on just the holographic part, then use straightforward video techniques to optically anchor it to an object in the room (making the framerate of the hologram less important to the immersiveness experience). In other words, with ML, you can look back & forth, and the scene around you doesn't 'slosh' because you're seeing the room itself directly, and the hologram just appears to wobble slightly. Since the BIG detail (the room) doesn't lag, your brain is more tolerant of the hologram's quirks.
Virtual/Augmented/Mixed Reality is now approaching its "Roomba" stage... finding ways to do useful things with presently-viable technology. We're a LONG way from general-purpose robotic servants (or a Star Trek Holodeck) & still have to hand-scrub behind the toilet, but at least the cat's daily furballs & tracked litter get cleaned with a button-press (and the equivalent of child-proofing the room to accommodate the robot's limits).
As far as the overwhelming majority of Americans are concerned, cars ARE king, and pedestrians are the intruders. You can debate whether that's good or bad, but the fact is, even people who LIVE in the urban cores of big cities almost always own cars unless they're prevented from doing so by virtue of poverty (or they use cabs and Uber SO OFTEN, the distinction between owning and not-owning a car is almost a matter of abstract semantics).
From the standpoint of that overwhelming majority, transit has actual MERIT only if it promises (truthfully or not) to reduce traffic congestion, and any form of transit that's likely (or practically GUARANTEED) to make traffic congestion worse is politically intolerable.
The utter failure of transit planners to come to terms with this reality is why Miami voters feel like victims of an elaborate bait & switch scheme every fucking time we get tricked into voting to increase transit funding. The county transit agency runs a multi-month advertising and public relations campaign that makes it look like they're planning to extend Metrorail down to Homestead, up to Broward, to South Beach, and out to West Dade... then six months later, quietly spends all the money on buses.
Fact: if Miami-Dade Transit had been 100% honest about how they intended to spend new transit funds (eg, on buses), voters would NEVER have agreed to either of the past two transit tax referendums. They would have gone down in flames and been defeated SO BADLY on election day, half the transit department's planning staff would have probably been laid off the following day in retribution.
Making matters worse, Miami's transit agency has an absolute KNACK for taking a good idea, then turning it into a proposal that's so fundamentally-flawed and awful, even people who SUPPORT building new rapid transit lines can't stomach it.
Specific examples I remember from recent years:
* The "East-West" Metrorail line, proposed to extend Metrorail from its current orange-line terminus at Miami International Airport west along State Road 836 (more or less) to Dolphin Mall, the south to Florida International University. Slam-dunk, right? Wrong. At one point (around 2007), they drew up fairly specific plans that showed where they planned to build stations. Let's just say that they literally couldn't have come up with worse proposed locations if they'd explicitly TRIED to achieve that as their primary design goal. Take the proposed station "at NW 87th Avenue". OK, solid choice... a square mile of medium-density office park to the northeast, a square mile of medium-density apartments and condos to the southeast, another square mile of medium-density apartments & condos to the southwest, and a HUGE lake with no room to build anything meaningful within walking distance to the northwest. Guess where they proposed putting the station? Yep... the northwest side of 836 and 87th Avenue. Literally ANYTHING of pedestrian interest would have been at least a half-mile walk away from the station. You could give Miami-Dade Transit a barrel full of fish & a shotgun, and they'd manage to explosively amputate their own foot. Any HALFWIT could look at Google Maps & realize that the sane place to put the station would be above the CSX railroad tracks, between NW 87th Avenue and NW 82nd Avenue (which would put much of the area to the north and south of the station within direct, easy walking distance, including the residential area south of 836 since 82nd Avenue now connects on both sides (it didn't back in 2007, but the reconstruction of SR836 was ALREADY underway, and if MDTA planners DIDN'T know about it, they sure as fucking hell SHOULD have).
* An EVEN WORSE plan to extend Metrorail west along SW 8th Street instead of 836... a route that would have not only managed to completely miss two huge malls (International Mall & Dolphin Mall) by 2 miles (as opposed to putting stations on the edge of their respective parking lots), but would have put two or three $50-100 million stations within conveni
The US is huge, but the population isn't all evenly-distributed. Look at a night satellite image of the US... the lights will show you where most Americans REALLY live.
Take Florida... vast & sprawling, but ~90% of Floridians live within 10 miles of I-95, I-4, or I-75. Ditto for Georgia, where ~80% of the state lives in Atlanta or Macon.
If "prime" property is used as a surface parking lot, its "prime-ness" is more a case of wishful thinking. Most new skyscrapers (the TRULY prime property) have their parking sandwiched between ground-floor retail & rooftop resort-like park/garden in the 10-20 story pedestal below the tower itself. One of the ironies of downtown Miami is that an average tower built within the past 20 years actually has more parking spaces per leasable square foot than Sawgrass Mills (a huge outlet mall on the Everglades-fringe of Fort Lauderdale). Planners *hate* it because they think pedestal garages are ugly, but they do mostly solve the parking problem without wasting surface/roof area.
This is why Musk's tunnel-boring is so important. Transit often makes gridlock on city streets WORSE... unless it has its own dedicated right of way. Tunnels are the least-objectionable way to do that, but traditional methods are just too expensive to do it large-scale.
With cheap tunneling, you CAN pitch transit to NIMBYs by saying, "it might reduce gridlock, and at LEAST won't make it even worse than it already is". You can't say that honestly about buses, streetcars, etc.
It's a shame the ADA made future transit projects based on overhead suspended cables impossible (no way to provide a 3-foot wheelchair-accessible egress path from vehicles... floor hatches with unrolled ladders to climb down aren't legal anymore), because it eliminated potential cheap solutions that could literally run overhead with widely-spaced support towers (especially as a way to increase the "reach" of subway stations by a mile or two perpendicular to the main line).
Musk's subways can potentially do that. You could take a subway station that was built in a less-than-ideal location (with a major trip-generator a mile away), and build an automated mini-subway a-la-Musk to shuttle people between that destination and the subway.
Illustrative example: downtown Miami. When Metrorail opened 35 years ago, the "downtown" station was a good half-mile west of what most of what Miamians considered "downtown", and the station in Brickell (Miami's original financial district) was ~2/3 mile from all the tall buildings. The elevated PeopleMover ("Metromover") VASTLY increased the reach of both stations & made most of downtown Miami accessible via Metrorail. Without Metrorail, Metromover would have been a useless toy. Without Metromover, 80-90% of the people who voluntarily used Metrorail (despite owning cars) to commute to downtown Miami in the 80s and 90s would have driven instead.
Modern front-loading washers really DO beat up on their passive components a lot harder than old ones. Old washing machines mostly rotated in one direction (or paused between changing directions), and had a fairly limited repertoire of motions. New ones, with digitally-controlled motors capable of acting like huge stepper motors constantly make radical changes to both direction and speed put ENORMOUSLY more stress on things like bearings, as well as the motors themselves. They also generally spin a LOT faster than old ones ever did, which doesn't just stress out the bearings and springs from increased use... it ALSO increases the likelihood that a slightly-unbalanced load will beat up on those components as it tries to accelerate up to its target rotation speed.
Seriously, just WATCH a modern front-load washer go through its motions in a typical load, then go to a laundromat & compare it to the typical motions of a 5-10 year old commercial front-loading washer. It's almost a miracle that home washing machines are capable of lasting for FIVE years without breaking catastrophically, given the forces they're subjected to when operating.
Combine the increased demands put upon their components with a general trend towards cost-reduction, and the breakage rate of modern washers is no great surprise.
Not quite (re "they all use AGPS").
There are several things that need to happen in order to get a GPS fix:
1. You need the satellite ephemera data. Android and IOS phones normally download this over the internet, because it's just a chunk of metadata. It's technically available directly from the satellites, but it takes about 10 minutes to receive a complete copy (the satellites only send a few bits at a time, interleaved between their timestamped beacon data). That's why old GPS devices took so long to get their first fix after powerup, and why they didn't take nearly as long if you kept them continuously powered on and someplace where they could get a clear signal.
2. You need a radio that can receive the satellite broadcasts. The silicon needed to do this first appeared around 15 years ago. The satellite broadcasts themselves are little more than timestamped beacons.
3. You need a way to correlate the timestamped beacon data with the ephemera data. This is where AGPS originally came in. To cut costs, the first phones with "GPS" worked by putting the radio receiver on the phone, but offloading the number-crunching to a remote server. Along the way, "AGPS" did other things that expedited the location lookup... if you know your APPROXIMATE location, it's a lot quicker to calculate your GPS-derived location than it is if you literally have no idea where you are. The more precisely you can approximate your location, the faster and easier it will be to calculate the real location. Since towers were at fixed, known locations, AGPS could say, "Hey, here's the radio data, and we know the user is somewhere within a few miles of (lat,lon) by virtue of using {some specific tower}".
Nowadays, what happens is more like this:
1. You reboot your phone, and it uses the internet to grab a fresh copy of the ephemera data.
2. Your phone knows which tower it's using. It does another query over the internet, and finds out the latitude and longitude of the tower. If it sees two or more towers, it can narrow down your location to approximately 1km.
3. Using the data from steps 1 and 2 to expedite the process, it does its own number-crunching on the GPS data.
In theory, a typical Android phone is capable of getting a GPS fix 100% offline... but it doesn't always work due to poorly-implemented software. Some devices don't even TRY to operate completely offline... if they can't access the internet, their software doesn't even TRY to obtain a GPS fix "the old fashioned way". Other devices have buggy firmware... say, they might fetch a copy of the ephemera data at bootup, but might never try to update it any other time... so if you go for a few weeks without rebooting your phone, the ephemera data will eventually become "stale", and your phone's ability to get a GPS fix will get worse and worse until you finally reboot.
Google changed the rules again a couple of years ago when they added location services to Google Play Services for Android. As I understand it, Play Services attempts to use all sources of location data available to it... if the phone is offline, it should (in theory) be able to fall back to raw, offline GPS... but that will ONLY work if the vendor who sold the device and built its firmware actually IMPLEMENTED offline GPS properly. Because it's so hard to test rigorously outside of lab conditions, a lot of custom kernels (and quite a few STOCK kernels) fuck it up and don't quite get it right (or even try at all).
In any case, on a modern Android phone, the "GPS" radio (which, on newer phones, can almost always handle Glonass as well... and probably BeiDou and Galileo, though the latter two constellations aren't quite 100% functional yet, and I have yet to see a stock ROM that even TRIES to make use of either one) is just one of several sources of location data. A lot of people call the way modern Android phones do GPS "AGPS" because of how it fetches the ephemera data and tower locations over the internet, but strictly speaking, it's NOT the same thing that "AG
Newer cell phones sniff wi-fi beacons & use crowdsourced services that correlate wifi ssids & relative signal strength to locations.
Actual GPS works poorly, or not at all, indoors (besides LITERALLY next to a window), and accuracy is poor because what little signal you get has so much multipath interference.
Mostly, because Condorcet logic is really hard to explain to normal people, and even HARDER to pitch to the media as something that can be presented via soundbites and headlines. My hybrid scheme makes the first round a relatively low-key internal affair for the House of Representatives, then gives the media 3 candidates to talk about & handles the Speaker's election in a relatively straightforward (by American standards) manner beyond that.
The main benefit of using Condorcet rules for round 1, and having the entire vote for Speaker via secret ballot, is ensuring that whomever ends up winning as speaker probably WON'T be the preferred first choice of either party... and that any Speaker who pisses off too many Representatives by getting overly heavy-handed about enforcing party discipline won't be re-elected as Speaker.
There's a second reform I can think of that would severely limit the power of party leaders over individual representatives: whenever a bill is defeated, there's an immediate and automatic secret "no confidence" vote among representatives. If the Speaker loses the vote, a new Speaker is elected immediately (under the same rules as above), and the newly-elected Speaker is not bound by the previous Speaker's committee appointments or policies.
This would put a stop to Speakers who ram bills through the house with single-vote victories by putting the Speaker's position in EXTREME peril if he allows a vote to proceed without being REALLY confident of a solid victory. It might work once... or twice... but eventually, s/he's going to piss off one Representative too many, the bill will be defeated, and the Speaker will be defeated as well. Every Speaker would have to choose between loyalty to his/her party, and desire to keep being Speaker. The relative power of parties to dictate legislation would probably ebb and flow, but any party that pushed TOO hard to control its Speaker would find itself having to continually re-establish its power over new Speakers -- each of whom would be harder to control than the last.
One good way to partially fix the House of Representatives would be to change the way the Speaker is elected:
1. Allow Representatives to vote for ANY House member to be Speaker... by secret ballot, using a Condorcet method. No nominations, and members can't refuse if they win (otherwise, parties would just punish any representative from their party who accepted the position without their approval). Representatives can vote for as many or as few members as they like, and indicate different preferences for those they vote for (with everyone they don't vote for at all being treated as "last choice, with equal preference... so even someone who loyally supports his or her party's choice for Speaker would, at a minimum, have to vote for everyone in the party... assigning the party's choice as their #1 choice, everyone else in the party as their #2 choice, and everyone else as choice #3).
2. The top three candidates from step 1 run against each other, once again via secret ballot among House members. If one of them gets a simple majority, he/she's the new Speaker. Otherwise...
3. The top two candidates from step 2 have a run-off election (also by secret ballot). If one of them gets a simple majority, he/she's the new Speaker.
4. If, however, step 2 produces a result where the top candidate wins a plurality & the remaining two are tied, or if step 3 produces a tie, step 2 is repeated... but this time, under Condorcet rules (as per step 1).
Electing the Speaker this way wouldn't be likely to result in a Speaker who's from a party different from a majority of Representatives... but it WOULD effectively throw a monkey wrench into either Party's ability to enforce party discipline on Representatives, and quite probably result in the election of Speakers who are absolutely, positively NOT the first choice of the Party's own leadership. A Speaker who gets to be TOO heavy-handed about bringing representatives in line would be unlikely to win again, because he'd ultimately piss off too many members of his own party. By keeping the votes for Speaker secret, Representatives could freely vote against those who've pissed them off or antagonized them without fear of reprisal or punishment by the Party.
Be glad you don't have astigmatism, or you'd find yourself in the position of being unable to see clearly without glasses (or eventually, without multifocal lenses) at ANY distance, near OR far.
Ever since roughly age 40, I've felt like I need a binocular microscope to solder anything smaller than 100-mil pitch... and depending on the part & lighting, even 100-mil has left me feeling like I'm "soldering blind" half the time. I've gotten to where I need a magnifying glass just to tell the difference between red and orange bands on a resistor, even WITH bright high-CRI lighting.
Because, ya know, SOMEONE needs to be responsible for sitting on the ISS and tattling on American & European crew members who spend Friday night over on the Russian side sharing a newly-arrived bottle of quality vodka. Or maybe even taking it upon him/herself to legally enforce temperance in orbit by confiscating and destroying any contraband vodka that might arrive on a Soyuz.
I partially jest... but the sad thing is, I can almost see something that stupid and petty happening.
The problem isn't that the water is too warm to run the plant. The problem is that heating lake water to 100+ degrees (F) will kill just about everything that lives in the lake (except algae, which will bloom like crazy and suffocate whatever the higher water temperature didn't kill directly), so they throttled back the plant to avoid overheating the lake.
For many, 3 unrestricted movies for $9.95/mo is a WAY better deal than $9.95/mo + $3-7/movie + tons of additional annoying restrictions.
100 watts is NOTHING if you have a gaming laptop or mobile workstation w/desktop CPU. The GPU *alone* can suck down more than 100 watts, and even when sipping power, you'd be forced to choose between "running" ("limping?") and "charging".
It probably ALSO includes a photographic printer (for high-res, high-quality copies), which would increase the complexity and weight quite a bit. Even if it did Kodak-style B&W "activator" development, the print path would have been pretty complex since the goal was "push-button, print-comes-out" simplicity for users.
I remember those readers... even back in the early 90s, my university charged $1.00/page to print microfilm & microfiche, back when normal copiers were 10c-25c/page (sometimes, 5c-9c if you had a high-value copycard). The only bright spot was that you could sometimes fit 2 pages on one print (microfilm) or 2x2 pages (microfiche) since the prints themselves were so high-resolution.
Do browsers NOT cache static content anymore? Or, in copyright frenzy, do sites mark & tag shit like JQuery, webfonts, etc as the browser-equivalent of "copy-never" to explicitly prevent it from being cached (even within the same site, let alone sites linked to the same static content [like JQuery, Bootstrap, etc.])?
It seems like the kind of stuff that *used* to be aggressively-cached (across multiple sites, no less) by browsers like MSIE & Firefox now LITERALLY gets fetched over and over again).