Has anyone written a program that can take a.wav sample of a poor-quality c64/vic 20 tape (stretched tape, warbling output, drifting volume level, etc) and do modern DSP analysis on it to turn it into something the more mundane.wav->.tap converters can handle? I unfortunately don't have the background to actually implement such a beast, but from what I know about both DSP and the way signals were recorded to tape by a Datasette/C2N, it *should* be a somewhat straightforward exercise. In theory, it might even be possible for a software DSP to recognize "print through" artifacts and accommodate for them.
Going a step further, does there exist a guide to the tokenization scheme used by BASIC on the Vic/64, so if there end up being a couple of mangled/indeterminate bytes, the proper values can be guessed based upon their context and the values of the surrounding bytes (assuming it's not in the middle of something mostly context-free, like a DATA statement)?
I have a couple of Vic-20 programs I wrote back in middle school that I'd love to recover for nostalgia reasons, but I've never been able to get any of the.wav conversion software to take them because the average volume level's drift tends to exceed the high/low threshold. In other words, I can only get a semi-intact CRC-error-free dump of the header and first few bytes if I set the threshold to something like 1 or 2, but within a second or two of the main recording starting, its average volume level drifts up or down enough to put "low" above the beginning's "high" threshold, or "high" below the beginning's "low" threshold. And most of the programs I've seen aren't much more sophisticated than the C64/Vic 20's own tapeloading routines were. For example, they might use the two copies to determine an error, but they don't even try to use the good parts of both copies to produce a single error-free result. The original hardware/software didn't do it because there wasn't enough RAM... but modern software has no such limit;-)
Never, ever gonna happen. At least, not publicly and visibly during our lifetimes. NANPA has fantasized about 11 and 12 digit numbers for years, and they've still got opposition levels from just about everyone, from end users to elected officials, that makes IPv6 look ENTHUSIASTICALLY *EMBRACED* by comparison.
So, what's REALLY going to happen? NAT. Just like with internet service, 10-digit phone numbers will cease to represent specific telephones (or at least specific telephone lines) at a specific location, and morph into DNS-like hostnames that can be mapped to any underlying 11 or 12-digit phone number. The act of freeing up phone numbers traditionally used for bulk PBX service (those in the form xxx-1xx-xxxx) alone will create enough new 10-digit numbers to keep individual American & Canadian users happy for the next hundred years.
To wit... right now, a company like Microsoft might have a few thousand directly-dialable phone numbers correlating to receptionists, autoattendants, and employees who receive a LOT of incoming calls, but probably has at least 20k or 30k actual phones that have "phantom" numbers like I mentioned above simply because under the existing telco switch system, they MUST have A number. Transition the 10-digit numbers into quasi-DNS and map them transparently to new 12-digit numbers, and give all the "bulk" phones at the call centers new 12-digit numbers, and those 20k-30k 10-digit numbers formerly wasted on call center phones can now be assigned to new customers. Stir, rinse, and repeat about a million times around the United States (particularly in places like New York that are particularly business-dense), and the "shortage" of 10-digit numbers starts to look about as "urgent" as it did 5 years ago (after millions of phone numbers formerly used by second/third phones for dialup internet, faxes, beepers, etc were dumped back into the number pool as their original uses became moot).
At the end of the day, 10-digit phone numbers aren't going anywhere in America. They might end up becoming nothing more than convenient DNS aliases behind the scenes, but 25-50 years from now, Americans and Canadians are still going to be dialing numbers that look the same as they do now, regardless of how much NAT & DNS magic takes place behind the scenes to make it work.
> The problem isn't that big businesses have to pay to keep themselves out of trouble. > The problem is, what if this works as intended? Then ultimately, it will lower the relative risks > of the member tech giants. So relatively speaking, every other small and medium business will have > to pay a risk tax that does not exist for these large companies.
In other words, they're re-inventing ASCAP in a different industry. They'll offer to let those small companies join, for an extortionate annual fee, then eventually start suing everyone IT/Telcom-related who's not a member on the presumption that they couldn't possibly be doing business without infringing on SOMETHING the group owns. Then AT&T, Microsoft, Dell, and Sony will start another company with similar, but non-overlapping, patent portfolio, and metaphorically become BMG... and hit those same small companies up for a SECOND, equally-extortionate, annual fee.
> Why don't batteries work for VOIP from the cable company?
Because cable companies, unlike phone companies, aren't required to have backup power to run THEMSELVES. Or, as Comcast's reps eloquently put it after Hurricane Wilma, "Our crews follow FPL's." No power == no cable == no cable internet. Hurricane Wilma left my old neighborhood's power lines relatively unscathed, but destroyed our power substation, so we had no power for more than two weeks (Coral Gables... central Dade County). I never lost DSL local loop, but Comcast didn't get service restored to the area until the day after FPL did, and didn't get it restored to pre-hurricane problem-free levels for another week. Anyone with cable internet in the area was SOL unless they had a tetherable PDA phone.
Sadly, when I moved recently, I was assured by AT&T/BellSouth that DSL is available in my neighborhood... then told that I can't actually GET DSL right now because their DSLAM ports are maxed out (it's a remote DSLAM), and the bastards are too cheap to add more. To say I was pissed would be an understatement... I actually had DSL availability as an explicit contingency on my purchase offer, and was delighted to have it officially satisfied the next day by BellSouth's assurances that DSL exists in my neighborhood (never bothering to mention that "Exists" != "Available to new customers"). So for now, at least, I'm a captive Comcast customer:(
Um, in case you haven't heard, American cell phone customers pay for airtime on outgoing AND incoming calls.
10 years ago, I thought it was a Bad Thing. Now, I think it was ultimately a Good Thing. Why? Because the people who bore the costs were the cellco's customers, so we exerted direct pressure on our carriers to include more minutes of airtime and/or make nights/weekends/incoming calls (or at least the first minute) free. Nowadays, a typical American cellular customer pays about $50/month for service that, for all intents and purposes, is almost flat-rate. Contrast that with Europe, where the cost to call mobile numbers is set by faceless bureaucrats who are accountable mainly to the carriers themselves, and there's no incentive (or possibly even a legal way) for individual carriers to reduce the cost of incoming calls and use it as a competitive advantage over others. The net result is that a frugal person who rarely makes outgoing mobile calls can get much cheaper service in Europe, but Americans who pay 2-3x as much can basically use their phones all day (or at least all night/weekend) until the battery runs out with little risk of getting a big bill at the end of the month.
> but you know the truth: unintentional DoS from the sheer volume of spam out there,
Oh god, don't remind me. Up until about 3 years ago, I ran my own mail server (DSL, fixed IP, old PC). One of the things I did was enable SASL authentication for SMTP (which requires logging in with a username and password before outgoing mail will be accepted for relay). Within a matter of months, spammers around the world figured out that I had a live SMTP server running on port 25. SASL AUTH or not, more and more spammers kept hammering away trying (unsuccessfully) to relay. My router's NAT table started to periodically overflow (crashing the router's firmware), and the endless incoming requests effectively were like a constant denial of service attack.
In retrospect, I could have probably gotten away with changing the SMTP server to a different port, but I was so fed up with the experience I ended up leasing a dedicated server for $30/month... partly, because once I knew what to look for, I noticed that I was ALSO getting hit by a staggering number of incoming http requests for various exploit-related URIs. The exploits themselves didn't bother me (I was running Tomcat as a standalone server), but on more than a few occasions I was getting hit with more than a hundred bogus http requests per minute.
Once I had my ISP change my IP address to a new one, my throughput more or less tripled, because I was no longer being DOS'ed 24/7 by bots, spammers, and worms. I pity anyone who has to maintain a live web/mail server today. ~10 years ago (when I used to wear both admin and developer hats at work) keeping a Linux server running was no big deal, and any halfway intelligent developer could do a decent part-time job of it. It was analogous to private security guards trying to keep kids from skateboarding in downtown parking garages. Now, it's more like trying to safeguard a business from looting during a riot.
> Doubtless. But inferior, cost-free tools sometimes make better, commercial ones unsellable.
That's an interesting observation. One development tool that's outrageously expensive AND non-negotiably mandatory for anyone who writes apps that interact with an Oracle database somewhere is Toad. As far as I know, Toad is without free (as in beer AND liberty) peer in the Win32 universe. There USED to be a decent (not spectacular, but good enough to limp along with if management wouldn't cough up a kilobuck or so for Toad) alternative called ToRA, but Quest (Toad's owner) bought them up and effectively abolished the Win32 version. The Linux version still exists, though (thank ${deity}).
I *can* think of one specific area where non-free tools are overwhelmingly preferred over anything open-source: m68k embedded development. If you do professional m68k/coldfire development, you use CodeWarrior. Period. If I had to name the single worst mistake PalmSource made (and it's hard, because they made so many), it was the recklessly premature deprecation of CodeWarrior as Palm's official development platform. Cobalt didn't have zero developer interest or support because manufacturers weren't interested in it... it had zero developer interest because the only way to write native code for it was to use PODS... and PODS sucked. It probably wouldn't have sucked forever, but by sucking so badly at the point in time when PalmSource desperately needed developer support the most, it was the final nail in Cobalt's coffin.
There's a good reason, though, why good open-source IDEs ultimately triumph over even comparably-good or slightly-better commercial IDEs: freedom-as-in-liberty. You can argue forever whether IntelliJ is better or worse than Netbeans and/or Eclipse, but one thing is certain -- its version control support (or lack thereof) is an ideological decision of its developers. They happen to believe that version control should be handled externally. Unfortunately for Jetbrains, plenty of developers would rather have transparently-integrated CVS and Subversion support that "just works". Netbeans' core developers tried to go in the same direction (abolishing seamless integrated CVS support in favor of less-capable generic support), and quickly got beaten up by angry users who took matters into their own hands (always an option with open-source) and put it right back in, along with equally transparent support for Subversion. In the market for commercial software, all developers could do is bitch, and possibly refuse to buy future releases... hoping that the commercial software's vendor eventually gets a clue (and doesn't just blame falling revenues on piracy). In the OSS universe, end users (at least, the more motivated ones) can forcibly make changes on their own.
Assuming you could get someone like WildBlue to provision it & you had the necessary (Cisco?) hardware to actually DO the deep packet inspection and convoluted routing, try THIS on for size:
* one-way (downlink-only) satellite
* EV-DO, for uplink AND for latency-sensitive downlink
Here's how such a system might handle a browser request:
* User navigates to a web URL somewhere
* DNS lookup is routed entirely via EV-DO. It's not a lot of data, but delays at this point are going to really, really hurt badly.
* http request is made for what appears to be html, css, xml, javascript, or something else that's text-based. It's sent out via EV-DO, and routed so the return comes via EV-DO as well.
* The browser receives the page, parses it, and discovers it needs 47 images. The http requests for the 47 image files go out via EV-DO, but are routed so that the response comes via satellite. The requests are parallelized as massively as possible to get the requests in the distant web server's response pipeline as quickly as possible. There's a delay of about 2-3 seconds from the moment the browser makes the initial request, but once the data starts to arrive, the page completes loading almost instantly.
Things like email would use EV-DO both ways for SMTP, but might use EV-DO up and satellite down for POP3 and/or IMAP4 (particularly if multiple incoming emails can be downloaded in parallel). Games would use EV-DO both ways, as would things like VNC and RDP.
Of course, at that point, it would probably be cheaper and faster to just get a T1 & use directional Wi-Fi antennas to split the monthly cost and share it with a neighbor or two...
> I'm skeptical the typical viewer cares about quality above the minimum threshold of a poorly encoded dvd.
Typical viewers? Maybe, maybe not. Now ask somebody with a huge TV in a small living room, and they'll tell you that it ABSOLUTELY matters. If you're sitting 8 feet or less away from a 56+ inch TV, anything less than DVD picture quality looks like a moving wall of quivering LEGO blocks. HD video, however, looks great at that distance, and you can enjoy the same immersive ambiance you get from good seats at a movie theater.
A lot of people who think HD isn't a big deal have been influenced by the real-world behavior of natively 1080i60 TVs. There's a good reason why 720p60 doesn't look much better than 480p60 on a 1080i60 set... the TV almost always resamples it from 1280x720 to 1280x540, then shows each line twice. You end up with more horizontal detail, but only slightly more vertical detail than 480i60. By the same token, "real" 1080i60 doesn't look all that impressive on a 1080i60 set compared to 480p60, either, because slightly less than half of its vertical resolution has to be averaged away to keep the screen from flickering like a 1980s radar weather map on crack. Compare 480i60 to 480p60, 720p60, and 1080p24/30/60 on a natively 1080p60 set, and you WILL see dramatic improvements with each step up... though natively-interlaced 1080i60 content (ie, not from 24fps film) will almost ALWAYS look like crap on a natively 1080p60 TV, because affordable motion-vector adaptive deinterlacing just doesn't exist at 1920x1080 yet... and thanks to HDCP, today's homebrew crowd can't repurpose an old PC into a ghetto-fabulous deinterlacer and enjoy almost the same quality "legitimate" customers pay $10-20k to enjoy with dedicated video processors.
> AND WiMax is not available, AND satellite isn't possible, AND dial-up isn't available. > I think if you lived in an area that remote, Comcast cable being in the ground is kind of a laughable impossibility.
Name one city in North America where WiMax exists today as a commercial service.
Satellite has hellish latencies that are intolerable to begin with, and made even worse by the way the satellite routers actually package the packets for uplink.
Dialup? You can't be serious. You might as well argue that mp3 players and terrestrial FM are real competition for satellite radio.
I second the recommendations for both PEFI and avrfreaks.net (caution: it's.NET, not.COM).
The AVR is a great, well-behaved MCU that's relatively cheap, amazingly tolerant of abuse, and supported by a website that's probably the single best site of its kind. The folks at avrfreaks.net are incredibly helpful, even if your problems lie more with electronics than with the AVR itself. For lots of us, avrfreaks.net is our "home", regardless of whether the specific project we're working on actually involves an AVR.
Don't get too hung up on analog stuff. Digital circuits do a great job of hiding your design sins, especially if your circuit basically consists of a 5v-tolerant MCU driving a few peripherals directly or through open-drain FET drivers at speeds below 20MHz.
> From a programming perspective, how hard is it to examine a couple of fields - country, state and zip code. All of the tax data resides in a table.
Not quite. Remember, zipcodes don't necessarily align with the boundaries of tax authorities. I live in the City of Pembroke Pines, Florida... but technically, my zipcode and official mailing address is the CIty of Hollywood, Florida. Someone going by zipcode would assume I'm within the jurisdiction of Hollywood, not Pembroke Pines.
Wait. It gets worse. I can't name a specific zipcode, but I know there are at least a few in the Florida panhandle where people who physically live in Florida have zipcodes associated with Georgia or Alabama (or vice-versa), because the post office that serves them is on the other side of the state line.
And like others have pointed out, it's hard enough to keep track of what's taxable, and at what rate, in the specific jurisdiction where your store is physically located. Expecting someone to be able to figure out which jurisdiction(s) apply, and what the final tax rate is under the collective rules that apply within that jurisdiction, imposes an unreasonable burden.
Plus, even if you've figured out the percentage, it's not necessarily consistent. For example, in Miami-Dade County, the sales tax is 7%... except it's really 6% with two 1/2% surtaxes on the first $5,000 of purchases made within the county. Or maybe it's 6 1/2% with one 1/2% surtax on the first $5,000 of purchases. I'm not entirely sure, and I LIVED in that damn county for almost half my life before fleeing to the Promised Land (Broward County) a few weeks ago. Trying to keep track of niggling details like this would be a MAJOR challenge for big internet vendors like Amazon, and point-blank impossible (or at least require expensive assistance from a third party) for smaller companies.
> The lines belong to them, especially the fiber lines, so what fucking right does the government > have to tell a company it isn't allowed to use it's own property as it wants even though it is > not infringing on anyone's rights?
They might own the fiber, but the overwhelming majority of land it runs across isn't owned by them... it's owned by governmental entities, most of whom took the land from its original owners without compensation. Yeah, lots of fiber was laid along old railroad corridors... most of whose land was acquired from the government the exact same way a century earlier.
If you lay fiber across land you've purchased in open market transactions from willing sellers under freely-negotiated terms, I agree... nobody should have the right to dictate what you can do with your fiber. HOWEVER, if even a single millimeter of that fiber crosses public right of way, that land's owner ("the public") has every right to dictate the terms under which the use of that land is made available. The fact that it happens to affect the use of that fiber upstream and downstream from that hypothetical millimeter of land doesn't change things.
I'm not a favor of governmental micromanagement. But there ARE real-world limits to the number of separately-owned fibers that can co-exist and compete for the same customers. Even if there's physical room for 32 fibers from 32 companies running down every street in America, if there are TOO MANY companies sticking their hands into the bundle, screw-ups that affect OTHER fiber owners WILL HAPPEN. Not to mention that if the full cost of paying someone to run a fiber drop from Company N's fiber bundle to YOUR demarc has to be borne entirely by you (let's say $500 at the extreme cheap end, which assumes there's actually a buried conduit available for that company's fiber to use already in place), you're not going to be likely to switch to a different company just to save $5/month, because it would take almost a decade at that rate to save enough money just to pay off the initial cost of doing the hookup in the first place. NOBODY needs the bandwidth of 32 fibers to their house. In the grand scheme of things, it's more cost effective to just run 2-4 fibers that "last mile" to everyone, and leave the actual switching up to routers at the respective fiber owners' network operations centers.
Also, if the number of potential service providers is limited, at least one of them needs to be regulated into reliability -- because otherwise, in the long run, there WON'T BE reliable providers of service, because reliability is almost never the most profitable economic strategy, and a company that tried to compete based exclusively on it would end up bleeding away shareholder value until it got bought up by a company eager to use its assets "more efficiently" and join the race to the bottom with everyone else.
> the other point is in Canada and the US, the cable-co and the Tele-co's don't really compete > that much, they seem to do a dance around the borders but don't really enter the vital territory
In real life, I work for a major telco, so I feel a bit entitled to have an opinion on the matter (disclaimer: 95% of what I personally believe happens to directly contradict the views expressed by my employer).
As a practical matter, the market needs two broad classes of companies providing "last-mile" wired data service:
* Companies that are required to offer service that's "five nines" reliable, costs be damned. Five-nines service isn't "cost-effective" for investor-owned utilities. Not even at high tariffs, because there simply aren't enough customers for whom it's sufficiently compelling to offload the entire cost of five-nines reliability onto them. However, by requiring that they provide at least "three nines" reliability to even their least-valuable customer, the costs get spread around enough to let society as a whole benefit from it. A.K.A., "the Phone Company".
* Companies that are free to race to the bottom, and compete to offer the cheapest, least-reliable service possible to people who want lots of cheap flash & bling, even if it breaks down a lot. A.K.A., "the Cable Company".
If both categories of companies are required to allow anyone to use their infrastructure for "last-mile" access to their ultimate customers, at tariffs that can be left up to the market... but required to be transparent and available to all comers who satisfy the official qualifications for service at a given rate, two are probably enough... though I'd love to see a third competitor to compete with the incumbent bottom-feeder, as long as it's understood that the two bottom-feeders aren't allowed to have any common business ownership whatsoever. In urban areas, a second "regulated/reliable" competitor would be nice, too (but not economically viable out in BFE).
At the moment, my datacom utopia doesn't really exist. The closest it comes is probably the rules under which AT&T operates in Florida... and would have come if Broward County had prevailed over Comcast and required them to open their network to other last-mile customers. High-speed wireless will be an interesting monkey wrench thrown into the gears, because (in dense urban areas, at least) it has the potential to compete as a true bottom-feeder. It's not likely to be reliable in the official telco sense (though "mainstream" wireless providers like Sprint and Verizon will probably evolve into that role over time), but IS (to consumers, at least) comparable in both speed and capacity to DSL.
Satellite? Hard to classify. On one hand, it has Terrabytes per second of cheap unidirectional bandwidth with a marginal per-user cost of almost zero (but "astronomical" costs to serve that first customer). On the other, it will never, ever be suitable for anything interactive and time-sensitive unless the speed of light happens to change (highly unlikely). Iridium was a nifty idea (treating a huge constellation of moving satellites as line-of-sight cellular towers over a ~hundred-mile radius region), but ultimately required too much capital for anything short of a government with unlimited cash to burn (China, perhaps?) to make it feasible... partly, because there just aren't enough compelling customers out in the hinterlands to make servicing them at costs less than what they'd spend to just lay a few hundred miles of new fiber (or living with geosynchronous VSAT's latency) worth it.
Lots of well-intentioned people involved with the "Net Neutrality" debate have the right idea, but are fighting the wrong war. The IMPORTANT battle is for "Last-Mile" neutrality. If AT&T is allowed to provision 6.0m/512k DSL for their own subscribers, they should be EQUALLY required to allow independent ISPs to get local loop access to their OWN customers at the same speed... and should pay AT&T the exact same monthly provisioning charges that AT&T's "official" DSL ISP pays. Ditto, for cable and FIOS -- if Comcast's ISP is allowed to sell local-loop connectivity that's burstable to 10m/6m, and (more or less) bottoms out at 768k/128k, other ISPs should be allowed to pick up THEIR OWN customers' traffic at Comcast's NOC and handle things from there.
As long as independent ISPs are guaranteed the right to take over customers' internet traffic from the point it reaches the network operations center, and are guaranteed the right to obtain local-loop service under the same open, transparent, nondiscriminatory terms as the incumbent's own ISP, "net neutrality" will be moot for anyone who cares enough to choose an ISP whose traffic-shaping strategy is acceptable to them.
Attempting to legislate equality of results is a dangerous, ultimately futile, lost cause. Equality of access, on the other hand, is fairly straightforward to define, and tends to do a sufficiently good job to make 99% of the original supporters happy. There will always be an edge case that loudly falls through the cracks, but for the most part it works, with minimal cost and inconvenience to everyone else in the long run.
Hypothetical examples of access-equality:
Comcast builds a network operations center, connected to their main center via fiber, and sells rack space at open, published rates. Comcast itself is required to locate their "own" ISP there, or at least route its own ISP's traffic through there, to ensure that they can't send the other ISPs to "the back of the bus" if/when the initial fiber link to the NOC gets saturated and needs to be upgraded. Likewise, Comcast has to charge other ISPs the exact same rates for rack space, backup power, etc. that it charges its own ISP. And if the rates are outrageous, Comcast has no right to balk if a bunch of ISPs get together, mount a single router in a rack, and use it to forward all of their own traffic to a different off-site location via their own fiber that Comcast has no right to touch. Ditto, for AT&T and Verizon. Trust me, Verizon won't abandon FIOS, even if last-mile neutrality were forced upon them. They can't afford to. If THEY don't run fiber, the local cable company WILL. Without fiber, Verizon will just be "the phone company" forever... and we all know that landline telephony is an increasingly-profitable, growing industry (*cough* *cough* *cough*)...
I wish I had mod points today for 'informative', because the article points out something that's not intuitively obvious and explains an observation I've made for the past 2-3 years (the time since RGB LED theater lights started to become commonplace) -- why things illuminated by the "white" light from a RGB triad tend to look like crap. Specifically, the observation that if you shine "yellow" light produced by red and green LEDs onto a surface that's painted/colored yellow, the surface looks almost black, and if you shine "orange" light produced by red and green LEDs onto an orange surface, it appears to be bright red.
In retrospect, it makes sense, and I feel like kicking myself because I should have figured it out a long time ago from what I know about light bulbs and color-rendering index (CRI). The reason halogen lightbulbs have a high CRI (~98 or ~99 when running at full power, I believe, compared to midday cloudless summer sunlight's CRI of 100) is because they produce light comprised of a nearly infinite number of discrete wavelengths (or at least enough to look like it to the eyes). Likewise, the reason why colors look a little "off" under most CFL bulbs (though often better than incandescent bulbs running at reduced power) is because their CRI tends to be in the low 90s. They produce light that's the sum of a small (few dozen?) number of different wavelengths. If you look at it on a spectrometer, you see a bunch of little spikes with nothing in between. Old/cheap fluorescent bulbs look even worse, because they used fewer different phosphors and had bigger gaps between spikes.
This is relevant, because it makes it obvious why attempts to light a stage using ONLY theater lights comprised of red, green, and blue LEDs will produce less than spectacular (if not downright odd) results -- effectively, a light that produces "white" light from red, green, and blue LEDs is like a poor-quality low-CRI fluorescent light from the distant past... but worse. It's not a question of "temperature" (bluish, yellowish, etc), because your eyes can compensate for THAT (it only becomes a big deal with you're dealing with film, or putting natural and artificial light side by side where they can be visually compared). Ergo, if you wanted to make stage lamps capable of reproducing halfway natural-looking color, you'd need to combine -- at the BARE minimum -- not just red, blue, and green LEDs... but also yellow, and augment it with high-CRI halogen lights. Cyan, magenta, and orange LEDs probably wouldn't hurt, either. You'd use the halogen lights for general illumination, and use the LED lights to tip the color balance for effect (being aware that the more you dim the halogen and brighter you make the LEDs, the more the color quality of the light is going to deteriorate).
> I'd be happy for a cure for the cold personally.
It's called Pleconaril. The FDA wouldn't approve it, even though it apparently works nicely and reliably stops a cold in its tracks, because they weren't happy about people taking antiviral meds for a disease (almost) nobody actually dies from.
They're now trying to get it approved for some other disease. With a little luck, they'll succeed -- once it's stocked by drugstores, doctors can legally prescribe it "off label" for anything they like. Including (of course) a cold.
> What makes sense at this point is to either stick with DVD, which is fine, or buy a Blu-Ray player. > It does not make sense to buy an HD-DVD player at any price.
You're assuming that many people buying HD-DVD players TODAY are primarily motivated by desire to watch NEW movies in HD, as opposed to picking up a cool new disposable toy because it was cheap enough to say 'Fuck it' and buy just to enjoy the novelty of, and have a future player for HD camcorder videos burned to DVD+R media.
The fact that Blockbuster still has HD-DVDs available to rent helps. As long as they don't try to pull a Netflix and pre-emptively convert my queue to regular DVD on the perverse assumption that I care more about the movie itself than the fact that it's HD, I'll remain a Blockbuster subscriber. The moment I find a non-HD DVD from Blockbuster in the mailbox, and no option for my queue that says, "Delete queued movies no longer available on HD-DVD (or at least move them to the very end, and don't even THINK about sending one unless there's absolutely nothing else in my queue that's available on HD-DVD)", I'll do what I did before -- set the HD-DVR to record anything that looks interesting on a HD premium movie channel, and keep it around until I feel like watching it.
Furthermore, it'll be a cold, snowy July day in downtown Miami before I voluntarily spend more than a pittance on anything touched by the tentacles of Sony -- a company so thoroughly evil, they make the DVD Forum look saintly and benevolent by comparison.
> Videos turn into a slideshow on my 2ghz Turion running Ubuntu.
What video chipset do you have, and what driver are you using? I have a creaky old Dell D600 laptop with 1.6GHz Pentium M and Mobile Radeon (officially-obsolete 9000 series) that dual-boots BOTH Ubuntu (Feisty Fawn) and Vista Ultimate, and can (usually) play flash videos just fine as long as I don't OFFICIALLY make it "full-screen". For some bizarre reason known only to ATI, I can play videos in a window maximized to fill the entire screen, but if I try to abolish the window entirely and give the whole display over to the window, it seems to forget about all of its onboard hardware DCT acceleration and exhibits the behavior you describe.
My recommendation: if you're using the open-source driver, try the proprietary binary one. If you're using the proprietary binary one, try the open-source driver. In either case, try viewing the video as both "full-screen" and in a maximized regular window... Four permutations, at least one of which will probably give you MUCH better results than you're currently seeing.
In Miami, the effect of DST's end is profound and dramatic with respect to evening traffic -- it's several orders of magnitude worse during the winter, especially during November and December. Why? People hate to get home from work after dark. No, not everybody... but most do. When the sun doesn't set until 7 or 8pm, most people leave their offices between 4:30 and 5:30... but ENOUGH "stragglers" linger until 6 or 7 to keep the 5pm traffic surge just slightly below the 'gridlock' threshold. When November hits and DST ends, just about EVERYONE runs for the door by 5pm in a desperate attempt to beat the sun, causing MAJOR gridlock that's visibly worse than it was the previous week.
My biggest complaint with federal law regarding timezones is the fact that a state can choose to opt out of DST entirely, but can't choose to make it year-round. I think it's safe to say that if Florida had that option, the part that's currently in 'Eastern' time would quickly vote to spend the whole year in EDT/AST. Even on the darkest day of November or December, there would STILL be visible sunlight by 8am in Tallahassee (the northwesternmost major city in 'EST/EDT' Florida). As for the part of the state that's in Central time (ie, Pensacola metro area), I'd leave it up to them to decide whether they'd rather move to CDT/EST permanently, or do whatever Alabama does. If they decided they'd rather follow Alabama and be up to 2 hours behind Tallahassee, so be it. Personally, I suspect Alabama and Georgia would quickly follow Florida and adopt DST year-round, and the "Pensacola" issue would quickly become moot anyway.
If anything, what happened in New Zealand is a perfect illustration (along with FPL) of why investor-owned power companies aren't necessarily the best idea unless there are harsh financial penalties that automatically kick in to punish them for failures to provide service.
The problem with something like power infrastructure is that the most cost-effective solution for its nominal owner isn't necessarily in the best interest of its customers, even if it nominally saves them money, because the cost for individual customers to personally mitigate even a single "once-in-a-lifetime" large-scale breakdown ends up exceeding the total savings anyway. To the power company, being down for 2 or more weeks means the loss of ordinary revenue from a subset of their customers. To the customers, it means being effectively shut down and having ZERO revenue (or being completely miserable) unless they spend a substantial amount of money buying and feeding a generator (post-Wilma, I was spending ~$36/day on gas to keep the generator running).
The main reason why FPL is one of the most profitable (if not THE most profitable) investor-owned utilities in America is because they've completely mastered the art of externalizing their costs and shifting their burden almost entirely onto their own customers. FPL saves $20 million, but it ends up costing their customers five times as much in lost business and misery. Wall Street smiles, though, because the losses incurred by their customers don't ultimately impact FPL's next quarterly results.
The solution isn't necessarily for the government to take over the power grid, but rather to set firm performance standards, and fine the hell out of FPL when they fail to meet them to ensure that FPL *does* directly feel pain, and lots of it, whenever their customers are without power, to ensure that they LITERALLY spare no expense in making sure that nothing short of a nuclear bomb can take down Florida's power grid for any significant period of time.
It's sad, but thanks to FPL and our largely-complicit state legislature, Florida has the power grid of a minor rural village in a POOR third-world country. Name ANY other place in the developed world with the size, population, and average wealth of Florida where it would EVER be considered acceptable to have more than a hundred thousand customers without power for more than TWO WEEKS after a hurricane that barely left a dent in anything besides the power grid itself (Hurricane Wilma... 15 days, 17 hours without power... in Coral f***ing Gables, 2 miles directly south of MIA, right smack in the middle of urban Dade County, lest anyone think I'm talking about some distant exurb out in the 'glades...)
Compounding the problem is FPL's refusal to bury lines unless the host municipality provides them with a brand new 20 foot wide easement dedicated ENTIRELY to FPL. Remember the outrageous $100+ billion estimates FPL gave in post-Wilma press conferences when asked about the cost of burying power lines statewide? Most of the estimate was for easement acquisition via eminent domain. Why is it that power companies in Europe can dig microtunnels for power lines a few feet below the streets in ancient city centers without disturbing a single cobblestone, but it's somehow impossible for FPL to bury power lines below a 10 foot wide pre-existing grassy easement running through people's back yards?
FPL is the worst excuse for a power company in America.
Re:Darling of the community!
on
Is AMD Dead Yet?
·
· Score: 4, Interesting
Back in the Netburst era, you didn't HAVE to be an AMD fanboy to know that the Pentium 4 totally sucked in every meaningful way compared to anything by AMD. Now that Intel's rejoined the rest of the sane universe, it's not as clear-cut anymore. They're still (usually) a little cheaper than Intel, but it's harder to draw a clear line between them and definitively say one is better than the other. Personally, I still tend to favor AMD, but if I were shopping for a notebook and one with core2duo happened to be massively on sale that particular week, I wouldn't avoid it like the black death the way I WOULD have bent over backwards to avoid the wretched quasi-mobile version of the Pentium 4.
Allegedly, a large-scale EMP blast could destroy just about every semiconductor on earth. If Intel or AMD were instantly reduced to making PNP and NPN transistors and trying to rebuild their entire chip portfolios from scratch -- with notes (hopefully in hardcopy and not only in now-unreadable digital form), how many years would it take just to regain the technology we had in 1970? Would we go through a decade of reinventing and using the 74xxx discrete logic chips (but maybe this time going straight to surface mounted chips?), or would we mostly skip over that era and go straight for CPLDs and FPGAs? Or TRY to skip over them and go straight to CPLDs & FPGAs, fail miserably (or price them out of the market), and end up wasting 5 years on a temporary technological/economic dead end?
Ditto, for CPUs. Would we even screw around with reinventing the Z80 & 6502? Try building a 680x0 with twenty thousand discrete chips? Go straight for ARM? Would the race be on to regain 2 gigahertz, or would multicore chips hit post-apocalypse computers before speeds even made it back up to 100MHz again?
Slashdot Mirror
Has anyone written a program that can take a .wav sample of a poor-quality c64/vic 20 tape (stretched tape, warbling output, drifting volume level, etc) and do modern DSP analysis on it to turn it into something the more mundane .wav->.tap converters can handle? I unfortunately don't have the background to actually implement such a beast, but from what I know about both DSP and the way signals were recorded to tape by a Datasette/C2N, it *should* be a somewhat straightforward exercise. In theory, it might even be possible for a software DSP to recognize "print through" artifacts and accommodate for them.
Going a step further, does there exist a guide to the tokenization scheme used by BASIC on the Vic/64, so if there end up being a couple of mangled/indeterminate bytes, the proper values can be guessed based upon their context and the values of the surrounding bytes (assuming it's not in the middle of something mostly context-free, like a DATA statement)?
I have a couple of Vic-20 programs I wrote back in middle school that I'd love to recover for nostalgia reasons, but I've never been able to get any of the .wav conversion software to take them because the average volume level's drift tends to exceed the high/low threshold. In other words, I can only get a semi-intact CRC-error-free dump of the header and first few bytes if I set the threshold to something like 1 or 2, but within a second or two of the main recording starting, its average volume level drifts up or down enough to put "low" above the beginning's "high" threshold, or "high" below the beginning's "low" threshold. And most of the programs I've seen aren't much more sophisticated than the C64/Vic 20's own tapeloading routines were. For example, they might use the two copies to determine an error, but they don't even try to use the good parts of both copies to produce a single error-free result. The original hardware/software didn't do it because there wasn't enough RAM... but modern software has no such limit ;-)
Never, ever gonna happen. At least, not publicly and visibly during our lifetimes. NANPA has fantasized about 11 and 12 digit numbers for years, and they've still got opposition levels from just about everyone, from end users to elected officials, that makes IPv6 look ENTHUSIASTICALLY *EMBRACED* by comparison.
So, what's REALLY going to happen? NAT. Just like with internet service, 10-digit phone numbers will cease to represent specific telephones (or at least specific telephone lines) at a specific location, and morph into DNS-like hostnames that can be mapped to any underlying 11 or 12-digit phone number. The act of freeing up phone numbers traditionally used for bulk PBX service (those in the form xxx-1xx-xxxx) alone will create enough new 10-digit numbers to keep individual American & Canadian users happy for the next hundred years.
To wit... right now, a company like Microsoft might have a few thousand directly-dialable phone numbers correlating to receptionists, autoattendants, and employees who receive a LOT of incoming calls, but probably has at least 20k or 30k actual phones that have "phantom" numbers like I mentioned above simply because under the existing telco switch system, they MUST have A number. Transition the 10-digit numbers into quasi-DNS and map them transparently to new 12-digit numbers, and give all the "bulk" phones at the call centers new 12-digit numbers, and those 20k-30k 10-digit numbers formerly wasted on call center phones can now be assigned to new customers. Stir, rinse, and repeat about a million times around the United States (particularly in places like New York that are particularly business-dense), and the "shortage" of 10-digit numbers starts to look about as "urgent" as it did 5 years ago (after millions of phone numbers formerly used by second/third phones for dialup internet, faxes, beepers, etc were dumped back into the number pool as their original uses became moot).
At the end of the day, 10-digit phone numbers aren't going anywhere in America. They might end up becoming nothing more than convenient DNS aliases behind the scenes, but 25-50 years from now, Americans and Canadians are still going to be dialing numbers that look the same as they do now, regardless of how much NAT & DNS magic takes place behind the scenes to make it work.
> The problem isn't that big businesses have to pay to keep themselves out of trouble.
> The problem is, what if this works as intended? Then ultimately, it will lower the relative risks
> of the member tech giants. So relatively speaking, every other small and medium business will have
> to pay a risk tax that does not exist for these large companies.
In other words, they're re-inventing ASCAP in a different industry. They'll offer to let those small companies join, for an extortionate annual fee, then eventually start suing everyone IT/Telcom-related who's not a member on the presumption that they couldn't possibly be doing business without infringing on SOMETHING the group owns. Then AT&T, Microsoft, Dell, and Sony will start another company with similar, but non-overlapping, patent portfolio, and metaphorically become BMG... and hit those same small companies up for a SECOND, equally-extortionate, annual fee.
> Why don't batteries work for VOIP from the cable company?
Because cable companies, unlike phone companies, aren't required to have backup power to run THEMSELVES. Or, as Comcast's reps eloquently put it after Hurricane Wilma, "Our crews follow FPL's." No power == no cable == no cable internet. Hurricane Wilma left my old neighborhood's power lines relatively unscathed, but destroyed our power substation, so we had no power for more than two weeks (Coral Gables... central Dade County). I never lost DSL local loop, but Comcast didn't get service restored to the area until the day after FPL did, and didn't get it restored to pre-hurricane problem-free levels for another week. Anyone with cable internet in the area was SOL unless they had a tetherable PDA phone.
Sadly, when I moved recently, I was assured by AT&T/BellSouth that DSL is available in my neighborhood... then told that I can't actually GET DSL right now because their DSLAM ports are maxed out (it's a remote DSLAM), and the bastards are too cheap to add more. To say I was pissed would be an understatement... I actually had DSL availability as an explicit contingency on my purchase offer, and was delighted to have it officially satisfied the next day by BellSouth's assurances that DSL exists in my neighborhood (never bothering to mention that "Exists" != "Available to new customers"). So for now, at least, I'm a captive Comcast customer :(
You don't even need the external USB enclosure. Use 'em with a USB-to-IDE cable, and treat the bare drives like a stack of backup tapes.
Um, in case you haven't heard, American cell phone customers pay for airtime on outgoing AND incoming calls.
10 years ago, I thought it was a Bad Thing. Now, I think it was ultimately a Good Thing. Why? Because the people who bore the costs were the cellco's customers, so we exerted direct pressure on our carriers to include more minutes of airtime and/or make nights/weekends/incoming calls (or at least the first minute) free. Nowadays, a typical American cellular customer pays about $50/month for service that, for all intents and purposes, is almost flat-rate. Contrast that with Europe, where the cost to call mobile numbers is set by faceless bureaucrats who are accountable mainly to the carriers themselves, and there's no incentive (or possibly even a legal way) for individual carriers to reduce the cost of incoming calls and use it as a competitive advantage over others. The net result is that a frugal person who rarely makes outgoing mobile calls can get much cheaper service in Europe, but Americans who pay 2-3x as much can basically use their phones all day (or at least all night/weekend) until the battery runs out with little risk of getting a big bill at the end of the month.
> but you know the truth: unintentional DoS from the sheer volume of spam out there,
Oh god, don't remind me. Up until about 3 years ago, I ran my own mail server (DSL, fixed IP, old PC). One of the things I did was enable SASL authentication for SMTP (which requires logging in with a username and password before outgoing mail will be accepted for relay). Within a matter of months, spammers around the world figured out that I had a live SMTP server running on port 25. SASL AUTH or not, more and more spammers kept hammering away trying (unsuccessfully) to relay. My router's NAT table started to periodically overflow (crashing the router's firmware), and the endless incoming requests effectively were like a constant denial of service attack.
In retrospect, I could have probably gotten away with changing the SMTP server to a different port, but I was so fed up with the experience I ended up leasing a dedicated server for $30/month... partly, because once I knew what to look for, I noticed that I was ALSO getting hit by a staggering number of incoming http requests for various exploit-related URIs. The exploits themselves didn't bother me (I was running Tomcat as a standalone server), but on more than a few occasions I was getting hit with more than a hundred bogus http requests per minute.
Once I had my ISP change my IP address to a new one, my throughput more or less tripled, because I was no longer being DOS'ed 24/7 by bots, spammers, and worms. I pity anyone who has to maintain a live web/mail server today. ~10 years ago (when I used to wear both admin and developer hats at work) keeping a Linux server running was no big deal, and any halfway intelligent developer could do a decent part-time job of it. It was analogous to private security guards trying to keep kids from skateboarding in downtown parking garages. Now, it's more like trying to safeguard a business from looting during a riot.
> Doubtless. But inferior, cost-free tools sometimes make better, commercial ones unsellable.
That's an interesting observation. One development tool that's outrageously expensive AND non-negotiably mandatory for anyone who writes apps that interact with an Oracle database somewhere is Toad. As far as I know, Toad is without free (as in beer AND liberty) peer in the Win32 universe. There USED to be a decent (not spectacular, but good enough to limp along with if management wouldn't cough up a kilobuck or so for Toad) alternative called ToRA, but Quest (Toad's owner) bought them up and effectively abolished the Win32 version. The Linux version still exists, though (thank ${deity}).
I *can* think of one specific area where non-free tools are overwhelmingly preferred over anything open-source: m68k embedded development. If you do professional m68k/coldfire development, you use CodeWarrior. Period. If I had to name the single worst mistake PalmSource made (and it's hard, because they made so many), it was the recklessly premature deprecation of CodeWarrior as Palm's official development platform. Cobalt didn't have zero developer interest or support because manufacturers weren't interested in it... it had zero developer interest because the only way to write native code for it was to use PODS... and PODS sucked. It probably wouldn't have sucked forever, but by sucking so badly at the point in time when PalmSource desperately needed developer support the most, it was the final nail in Cobalt's coffin.
There's a good reason, though, why good open-source IDEs ultimately triumph over even comparably-good or slightly-better commercial IDEs: freedom-as-in-liberty. You can argue forever whether IntelliJ is better or worse than Netbeans and/or Eclipse, but one thing is certain -- its version control support (or lack thereof) is an ideological decision of its developers. They happen to believe that version control should be handled externally. Unfortunately for Jetbrains, plenty of developers would rather have transparently-integrated CVS and Subversion support that "just works". Netbeans' core developers tried to go in the same direction (abolishing seamless integrated CVS support in favor of less-capable generic support), and quickly got beaten up by angry users who took matters into their own hands (always an option with open-source) and put it right back in, along with equally transparent support for Subversion. In the market for commercial software, all developers could do is bitch, and possibly refuse to buy future releases... hoping that the commercial software's vendor eventually gets a clue (and doesn't just blame falling revenues on piracy). In the OSS universe, end users (at least, the more motivated ones) can forcibly make changes on their own.
Assuming you could get someone like WildBlue to provision it & you had the necessary (Cisco?) hardware to actually DO the deep packet inspection and convoluted routing, try THIS on for size:
* one-way (downlink-only) satellite
* EV-DO, for uplink AND for latency-sensitive downlink
Here's how such a system might handle a browser request:
* User navigates to a web URL somewhere
* DNS lookup is routed entirely via EV-DO. It's not a lot of data, but delays at this point are going to really, really hurt badly.
* http request is made for what appears to be html, css, xml, javascript, or something else that's text-based. It's sent out via EV-DO, and routed so the return comes via EV-DO as well.
* The browser receives the page, parses it, and discovers it needs 47 images. The http requests for the 47 image files go out via EV-DO, but are routed so that the response comes via satellite. The requests are parallelized as massively as possible to get the requests in the distant web server's response pipeline as quickly as possible. There's a delay of about 2-3 seconds from the moment the browser makes the initial request, but once the data starts to arrive, the page completes loading almost instantly.
Things like email would use EV-DO both ways for SMTP, but might use EV-DO up and satellite down for POP3 and/or IMAP4 (particularly if multiple incoming emails can be downloaded in parallel). Games would use EV-DO both ways, as would things like VNC and RDP.
Of course, at that point, it would probably be cheaper and faster to just get a T1 & use directional Wi-Fi antennas to split the monthly cost and share it with a neighbor or two...
> I'm skeptical the typical viewer cares about quality above the minimum threshold of a poorly encoded dvd.
Typical viewers? Maybe, maybe not. Now ask somebody with a huge TV in a small living room, and they'll tell you that it ABSOLUTELY matters. If you're sitting 8 feet or less away from a 56+ inch TV, anything less than DVD picture quality looks like a moving wall of quivering LEGO blocks. HD video, however, looks great at that distance, and you can enjoy the same immersive ambiance you get from good seats at a movie theater.
A lot of people who think HD isn't a big deal have been influenced by the real-world behavior of natively 1080i60 TVs. There's a good reason why 720p60 doesn't look much better than 480p60 on a 1080i60 set... the TV almost always resamples it from 1280x720 to 1280x540, then shows each line twice. You end up with more horizontal detail, but only slightly more vertical detail than 480i60. By the same token, "real" 1080i60 doesn't look all that impressive on a 1080i60 set compared to 480p60, either, because slightly less than half of its vertical resolution has to be averaged away to keep the screen from flickering like a 1980s radar weather map on crack. Compare 480i60 to 480p60, 720p60, and 1080p24/30/60 on a natively 1080p60 set, and you WILL see dramatic improvements with each step up... though natively-interlaced 1080i60 content (ie, not from 24fps film) will almost ALWAYS look like crap on a natively 1080p60 TV, because affordable motion-vector adaptive deinterlacing just doesn't exist at 1920x1080 yet... and thanks to HDCP, today's homebrew crowd can't repurpose an old PC into a ghetto-fabulous deinterlacer and enjoy almost the same quality "legitimate" customers pay $10-20k to enjoy with dedicated video processors.
> AND WiMax is not available, AND satellite isn't possible, AND dial-up isn't available.
> I think if you lived in an area that remote, Comcast cable being in the ground is kind of a laughable impossibility.
Name one city in North America where WiMax exists today as a commercial service.
Satellite has hellish latencies that are intolerable to begin with, and made even worse by the way the satellite routers actually package the packets for uplink.
Dialup? You can't be serious. You might as well argue that mp3 players and terrestrial FM are real competition for satellite radio.
I second the recommendations for both PEFI and avrfreaks.net (caution: it's .NET, not .COM).
The AVR is a great, well-behaved MCU that's relatively cheap, amazingly tolerant of abuse, and supported by a website that's probably the single best site of its kind. The folks at avrfreaks.net are incredibly helpful, even if your problems lie more with electronics than with the AVR itself. For lots of us, avrfreaks.net is our "home", regardless of whether the specific project we're working on actually involves an AVR.
Don't get too hung up on analog stuff. Digital circuits do a great job of hiding your design sins, especially if your circuit basically consists of a 5v-tolerant MCU driving a few peripherals directly or through open-drain FET drivers at speeds below 20MHz.
> From a programming perspective, how hard is it to examine a couple of fields - country, state and zip code. All of the tax data resides in a table.
Not quite. Remember, zipcodes don't necessarily align with the boundaries of tax authorities. I live in the City of Pembroke Pines, Florida... but technically, my zipcode and official mailing address is the CIty of Hollywood, Florida. Someone going by zipcode would assume I'm within the jurisdiction of Hollywood, not Pembroke Pines.
Wait. It gets worse. I can't name a specific zipcode, but I know there are at least a few in the Florida panhandle where people who physically live in Florida have zipcodes associated with Georgia or Alabama (or vice-versa), because the post office that serves them is on the other side of the state line.
And like others have pointed out, it's hard enough to keep track of what's taxable, and at what rate, in the specific jurisdiction where your store is physically located. Expecting someone to be able to figure out which jurisdiction(s) apply, and what the final tax rate is under the collective rules that apply within that jurisdiction, imposes an unreasonable burden.
Plus, even if you've figured out the percentage, it's not necessarily consistent. For example, in Miami-Dade County, the sales tax is 7%... except it's really 6% with two 1/2% surtaxes on the first $5,000 of purchases made within the county. Or maybe it's 6 1/2% with one 1/2% surtax on the first $5,000 of purchases. I'm not entirely sure, and I LIVED in that damn county for almost half my life before fleeing to the Promised Land (Broward County) a few weeks ago. Trying to keep track of niggling details like this would be a MAJOR challenge for big internet vendors like Amazon, and point-blank impossible (or at least require expensive assistance from a third party) for smaller companies.
> The lines belong to them, especially the fiber lines, so what fucking right does the government
> have to tell a company it isn't allowed to use it's own property as it wants even though it is
> not infringing on anyone's rights?
They might own the fiber, but the overwhelming majority of land it runs across isn't owned by them... it's owned by governmental entities, most of whom took the land from its original owners without compensation. Yeah, lots of fiber was laid along old railroad corridors... most of whose land was acquired from the government the exact same way a century earlier.
If you lay fiber across land you've purchased in open market transactions from willing sellers under freely-negotiated terms, I agree... nobody should have the right to dictate what you can do with your fiber. HOWEVER, if even a single millimeter of that fiber crosses public right of way, that land's owner ("the public") has every right to dictate the terms under which the use of that land is made available. The fact that it happens to affect the use of that fiber upstream and downstream from that hypothetical millimeter of land doesn't change things.
I'm not a favor of governmental micromanagement. But there ARE real-world limits to the number of separately-owned fibers that can co-exist and compete for the same customers. Even if there's physical room for 32 fibers from 32 companies running down every street in America, if there are TOO MANY companies sticking their hands into the bundle, screw-ups that affect OTHER fiber owners WILL HAPPEN. Not to mention that if the full cost of paying someone to run a fiber drop from Company N's fiber bundle to YOUR demarc has to be borne entirely by you (let's say $500 at the extreme cheap end, which assumes there's actually a buried conduit available for that company's fiber to use already in place), you're not going to be likely to switch to a different company just to save $5/month, because it would take almost a decade at that rate to save enough money just to pay off the initial cost of doing the hookup in the first place. NOBODY needs the bandwidth of 32 fibers to their house. In the grand scheme of things, it's more cost effective to just run 2-4 fibers that "last mile" to everyone, and leave the actual switching up to routers at the respective fiber owners' network operations centers.
Also, if the number of potential service providers is limited, at least one of them needs to be regulated into reliability -- because otherwise, in the long run, there WON'T BE reliable providers of service, because reliability is almost never the most profitable economic strategy, and a company that tried to compete based exclusively on it would end up bleeding away shareholder value until it got bought up by a company eager to use its assets "more efficiently" and join the race to the bottom with everyone else.
> the other point is in Canada and the US, the cable-co and the Tele-co's don't really compete
> that much, they seem to do a dance around the borders but don't really enter the vital territory
In real life, I work for a major telco, so I feel a bit entitled to have an opinion on the matter (disclaimer: 95% of what I personally believe happens to directly contradict the views expressed by my employer).
As a practical matter, the market needs two broad classes of companies providing "last-mile" wired data service:
* Companies that are required to offer service that's "five nines" reliable, costs be damned. Five-nines service isn't "cost-effective" for investor-owned utilities. Not even at high tariffs, because there simply aren't enough customers for whom it's sufficiently compelling to offload the entire cost of five-nines reliability onto them. However, by requiring that they provide at least "three nines" reliability to even their least-valuable customer, the costs get spread around enough to let society as a whole benefit from it. A.K.A., "the Phone Company".
* Companies that are free to race to the bottom, and compete to offer the cheapest, least-reliable service possible to people who want lots of cheap flash & bling, even if it breaks down a lot. A.K.A., "the Cable Company".
If both categories of companies are required to allow anyone to use their infrastructure for "last-mile" access to their ultimate customers, at tariffs that can be left up to the market... but required to be transparent and available to all comers who satisfy the official qualifications for service at a given rate, two are probably enough... though I'd love to see a third competitor to compete with the incumbent bottom-feeder, as long as it's understood that the two bottom-feeders aren't allowed to have any common business ownership whatsoever. In urban areas, a second "regulated/reliable" competitor would be nice, too (but not economically viable out in BFE).
At the moment, my datacom utopia doesn't really exist. The closest it comes is probably the rules under which AT&T operates in Florida... and would have come if Broward County had prevailed over Comcast and required them to open their network to other last-mile customers. High-speed wireless will be an interesting monkey wrench thrown into the gears, because (in dense urban areas, at least) it has the potential to compete as a true bottom-feeder. It's not likely to be reliable in the official telco sense (though "mainstream" wireless providers like Sprint and Verizon will probably evolve into that role over time), but IS (to consumers, at least) comparable in both speed and capacity to DSL.
Satellite? Hard to classify. On one hand, it has Terrabytes per second of cheap unidirectional bandwidth with a marginal per-user cost of almost zero (but "astronomical" costs to serve that first customer). On the other, it will never, ever be suitable for anything interactive and time-sensitive unless the speed of light happens to change (highly unlikely). Iridium was a nifty idea (treating a huge constellation of moving satellites as line-of-sight cellular towers over a ~hundred-mile radius region), but ultimately required too much capital for anything short of a government with unlimited cash to burn (China, perhaps?) to make it feasible... partly, because there just aren't enough compelling customers out in the hinterlands to make servicing them at costs less than what they'd spend to just lay a few hundred miles of new fiber (or living with geosynchronous VSAT's latency) worth it.
Lots of well-intentioned people involved with the "Net Neutrality" debate have the right idea, but are fighting the wrong war. The IMPORTANT battle is for "Last-Mile" neutrality. If AT&T is allowed to provision 6.0m/512k DSL for their own subscribers, they should be EQUALLY required to allow independent ISPs to get local loop access to their OWN customers at the same speed... and should pay AT&T the exact same monthly provisioning charges that AT&T's "official" DSL ISP pays. Ditto, for cable and FIOS -- if Comcast's ISP is allowed to sell local-loop connectivity that's burstable to 10m/6m, and (more or less) bottoms out at 768k/128k, other ISPs should be allowed to pick up THEIR OWN customers' traffic at Comcast's NOC and handle things from there.
As long as independent ISPs are guaranteed the right to take over customers' internet traffic from the point it reaches the network operations center, and are guaranteed the right to obtain local-loop service under the same open, transparent, nondiscriminatory terms as the incumbent's own ISP, "net neutrality" will be moot for anyone who cares enough to choose an ISP whose traffic-shaping strategy is acceptable to them.
Attempting to legislate equality of results is a dangerous, ultimately futile, lost cause. Equality of access, on the other hand, is fairly straightforward to define, and tends to do a sufficiently good job to make 99% of the original supporters happy. There will always be an edge case that loudly falls through the cracks, but for the most part it works, with minimal cost and inconvenience to everyone else in the long run.
Hypothetical examples of access-equality:
Comcast builds a network operations center, connected to their main center via fiber, and sells rack space at open, published rates. Comcast itself is required to locate their "own" ISP there, or at least route its own ISP's traffic through there, to ensure that they can't send the other ISPs to "the back of the bus" if/when the initial fiber link to the NOC gets saturated and needs to be upgraded. Likewise, Comcast has to charge other ISPs the exact same rates for rack space, backup power, etc. that it charges its own ISP. And if the rates are outrageous, Comcast has no right to balk if a bunch of ISPs get together, mount a single router in a rack, and use it to forward all of their own traffic to a different off-site location via their own fiber that Comcast has no right to touch. Ditto, for AT&T and Verizon. Trust me, Verizon won't abandon FIOS, even if last-mile neutrality were forced upon them. They can't afford to. If THEY don't run fiber, the local cable company WILL. Without fiber, Verizon will just be "the phone company" forever... and we all know that landline telephony is an increasingly-profitable, growing industry (*cough* *cough* *cough*)...
I wish I had mod points today for 'informative', because the article points out something that's not intuitively obvious and explains an observation I've made for the past 2-3 years (the time since RGB LED theater lights started to become commonplace) -- why things illuminated by the "white" light from a RGB triad tend to look like crap. Specifically, the observation that if you shine "yellow" light produced by red and green LEDs onto a surface that's painted/colored yellow, the surface looks almost black, and if you shine "orange" light produced by red and green LEDs onto an orange surface, it appears to be bright red.
In retrospect, it makes sense, and I feel like kicking myself because I should have figured it out a long time ago from what I know about light bulbs and color-rendering index (CRI). The reason halogen lightbulbs have a high CRI (~98 or ~99 when running at full power, I believe, compared to midday cloudless summer sunlight's CRI of 100) is because they produce light comprised of a nearly infinite number of discrete wavelengths (or at least enough to look like it to the eyes). Likewise, the reason why colors look a little "off" under most CFL bulbs (though often better than incandescent bulbs running at reduced power) is because their CRI tends to be in the low 90s. They produce light that's the sum of a small (few dozen?) number of different wavelengths. If you look at it on a spectrometer, you see a bunch of little spikes with nothing in between. Old/cheap fluorescent bulbs look even worse, because they used fewer different phosphors and had bigger gaps between spikes.
This is relevant, because it makes it obvious why attempts to light a stage using ONLY theater lights comprised of red, green, and blue LEDs will produce less than spectacular (if not downright odd) results -- effectively, a light that produces "white" light from red, green, and blue LEDs is like a poor-quality low-CRI fluorescent light from the distant past... but worse. It's not a question of "temperature" (bluish, yellowish, etc), because your eyes can compensate for THAT (it only becomes a big deal with you're dealing with film, or putting natural and artificial light side by side where they can be visually compared). Ergo, if you wanted to make stage lamps capable of reproducing halfway natural-looking color, you'd need to combine -- at the BARE minimum -- not just red, blue, and green LEDs... but also yellow, and augment it with high-CRI halogen lights. Cyan, magenta, and orange LEDs probably wouldn't hurt, either. You'd use the halogen lights for general illumination, and use the LED lights to tip the color balance for effect (being aware that the more you dim the halogen and brighter you make the LEDs, the more the color quality of the light is going to deteriorate).
> I'd be happy for a cure for the cold personally.
It's called Pleconaril. The FDA wouldn't approve it, even though it apparently works nicely and reliably stops a cold in its tracks, because they weren't happy about people taking antiviral meds for a disease (almost) nobody actually dies from.
They're now trying to get it approved for some other disease. With a little luck, they'll succeed -- once it's stocked by drugstores, doctors can legally prescribe it "off label" for anything they like. Including (of course) a cold.
> What makes sense at this point is to either stick with DVD, which is fine, or buy a Blu-Ray player.
> It does not make sense to buy an HD-DVD player at any price.
You're assuming that many people buying HD-DVD players TODAY are primarily motivated by desire to watch NEW movies in HD, as opposed to picking up a cool new disposable toy because it was cheap enough to say 'Fuck it' and buy just to enjoy the novelty of, and have a future player for HD camcorder videos burned to DVD+R media.
The fact that Blockbuster still has HD-DVDs available to rent helps. As long as they don't try to pull a Netflix and pre-emptively convert my queue to regular DVD on the perverse assumption that I care more about the movie itself than the fact that it's HD, I'll remain a Blockbuster subscriber. The moment I find a non-HD DVD from Blockbuster in the mailbox, and no option for my queue that says, "Delete queued movies no longer available on HD-DVD (or at least move them to the very end, and don't even THINK about sending one unless there's absolutely nothing else in my queue that's available on HD-DVD)", I'll do what I did before -- set the HD-DVR to record anything that looks interesting on a HD premium movie channel, and keep it around until I feel like watching it.
Furthermore, it'll be a cold, snowy July day in downtown Miami before I voluntarily spend more than a pittance on anything touched by the tentacles of Sony -- a company so thoroughly evil, they make the DVD Forum look saintly and benevolent by comparison.
---
"One down, Blu to go..."
> Videos turn into a slideshow on my 2ghz Turion running Ubuntu.
What video chipset do you have, and what driver are you using? I have a creaky old Dell D600 laptop with 1.6GHz Pentium M and Mobile Radeon (officially-obsolete 9000 series) that dual-boots BOTH Ubuntu (Feisty Fawn) and Vista Ultimate, and can (usually) play flash videos just fine as long as I don't OFFICIALLY make it "full-screen". For some bizarre reason known only to ATI, I can play videos in a window maximized to fill the entire screen, but if I try to abolish the window entirely and give the whole display over to the window, it seems to forget about all of its onboard hardware DCT acceleration and exhibits the behavior you describe.
My recommendation: if you're using the open-source driver, try the proprietary binary one. If you're using the proprietary binary one, try the open-source driver. In either case, try viewing the video as both "full-screen" and in a maximized regular window... Four permutations, at least one of which will probably give you MUCH better results than you're currently seeing.
In Miami, the effect of DST's end is profound and dramatic with respect to evening traffic -- it's several orders of magnitude worse during the winter, especially during November and December. Why? People hate to get home from work after dark. No, not everybody... but most do. When the sun doesn't set until 7 or 8pm, most people leave their offices between 4:30 and 5:30... but ENOUGH "stragglers" linger until 6 or 7 to keep the 5pm traffic surge just slightly below the 'gridlock' threshold. When November hits and DST ends, just about EVERYONE runs for the door by 5pm in a desperate attempt to beat the sun, causing MAJOR gridlock that's visibly worse than it was the previous week.
My biggest complaint with federal law regarding timezones is the fact that a state can choose to opt out of DST entirely, but can't choose to make it year-round. I think it's safe to say that if Florida had that option, the part that's currently in 'Eastern' time would quickly vote to spend the whole year in EDT/AST. Even on the darkest day of November or December, there would STILL be visible sunlight by 8am in Tallahassee (the northwesternmost major city in 'EST/EDT' Florida). As for the part of the state that's in Central time (ie, Pensacola metro area), I'd leave it up to them to decide whether they'd rather move to CDT/EST permanently, or do whatever Alabama does. If they decided they'd rather follow Alabama and be up to 2 hours behind Tallahassee, so be it. Personally, I suspect Alabama and Georgia would quickly follow Florida and adopt DST year-round, and the "Pensacola" issue would quickly become moot anyway.
If anything, what happened in New Zealand is a perfect illustration (along with FPL) of why investor-owned power companies aren't necessarily the best idea unless there are harsh financial penalties that automatically kick in to punish them for failures to provide service.
The problem with something like power infrastructure is that the most cost-effective solution for its nominal owner isn't necessarily in the best interest of its customers, even if it nominally saves them money, because the cost for individual customers to personally mitigate even a single "once-in-a-lifetime" large-scale breakdown ends up exceeding the total savings anyway. To the power company, being down for 2 or more weeks means the loss of ordinary revenue from a subset of their customers. To the customers, it means being effectively shut down and having ZERO revenue (or being completely miserable) unless they spend a substantial amount of money buying and feeding a generator (post-Wilma, I was spending ~$36/day on gas to keep the generator running).
The main reason why FPL is one of the most profitable (if not THE most profitable) investor-owned utilities in America is because they've completely mastered the art of externalizing their costs and shifting their burden almost entirely onto their own customers. FPL saves $20 million, but it ends up costing their customers five times as much in lost business and misery. Wall Street smiles, though, because the losses incurred by their customers don't ultimately impact FPL's next quarterly results.
The solution isn't necessarily for the government to take over the power grid, but rather to set firm performance standards, and fine the hell out of FPL when they fail to meet them to ensure that FPL *does* directly feel pain, and lots of it, whenever their customers are without power, to ensure that they LITERALLY spare no expense in making sure that nothing short of a nuclear bomb can take down Florida's power grid for any significant period of time.
It's sad, but thanks to FPL and our largely-complicit state legislature, Florida has the power grid of a minor rural village in a POOR third-world country. Name ANY other place in the developed world with the size, population, and average wealth of Florida where it would EVER be considered acceptable to have more than a hundred thousand customers without power for more than TWO WEEKS after a hurricane that barely left a dent in anything besides the power grid itself (Hurricane Wilma... 15 days, 17 hours without power... in Coral f***ing Gables, 2 miles directly south of MIA, right smack in the middle of urban Dade County, lest anyone think I'm talking about some distant exurb out in the 'glades...)
Compounding the problem is FPL's refusal to bury lines unless the host municipality provides them with a brand new 20 foot wide easement dedicated ENTIRELY to FPL. Remember the outrageous $100+ billion estimates FPL gave in post-Wilma press conferences when asked about the cost of burying power lines statewide? Most of the estimate was for easement acquisition via eminent domain. Why is it that power companies in Europe can dig microtunnels for power lines a few feet below the streets in ancient city centers without disturbing a single cobblestone, but it's somehow impossible for FPL to bury power lines below a 10 foot wide pre-existing grassy easement running through people's back yards?
FPL is the worst excuse for a power company in America.
Back in the Netburst era, you didn't HAVE to be an AMD fanboy to know that the Pentium 4 totally sucked in every meaningful way compared to anything by AMD. Now that Intel's rejoined the rest of the sane universe, it's not as clear-cut anymore. They're still (usually) a little cheaper than Intel, but it's harder to draw a clear line between them and definitively say one is better than the other. Personally, I still tend to favor AMD, but if I were shopping for a notebook and one with core2duo happened to be massively on sale that particular week, I wouldn't avoid it like the black death the way I WOULD have bent over backwards to avoid the wretched quasi-mobile version of the Pentium 4.
That brings up an interesting point.
Allegedly, a large-scale EMP blast could destroy just about every semiconductor on earth. If Intel or AMD were instantly reduced to making PNP and NPN transistors and trying to rebuild their entire chip portfolios from scratch -- with notes (hopefully in hardcopy and not only in now-unreadable digital form), how many years would it take just to regain the technology we had in 1970? Would we go through a decade of reinventing and using the 74xxx discrete logic chips (but maybe this time going straight to surface mounted chips?), or would we mostly skip over that era and go straight for CPLDs and FPGAs? Or TRY to skip over them and go straight to CPLDs & FPGAs, fail miserably (or price them out of the market), and end up wasting 5 years on a temporary technological/economic dead end?
Ditto, for CPUs. Would we even screw around with reinventing the Z80 & 6502? Try building a 680x0 with twenty thousand discrete chips? Go straight for ARM? Would the race be on to regain 2 gigahertz, or would multicore chips hit post-apocalypse computers before speeds even made it back up to 100MHz again?