AGP is a cheap hack. It was created because it is slightly cheaper than 66 MHz 64-bit PCI (actually, quite a bit cheaper).
The 3dfx cards used AGP AS 66 MHz, 64-bit PCI (not using the AGP part).
We're NOT using AGP the way it was intended. The goal of AGP was to let you eliminate graphics memory by creating a direct link to system memory. Intel's goal it to make everything dependant on the primary system (CPU), and as a result, putting it to memory requirements is second best.
Intel wants to take the money AWAY from component makes and move it into CPUs, mobos, chipsets, and other core features that they dominate.
AGP isn't faster than 66MHZ 64-bit PCI, it just a less complicated approach.
The spec for server PCI has been on the Alphas for about 4 or 5 years now? That should have been the solution for new graphics cards. Instead of the ugly AGP hack (and REAL cards don't take advantage of the AGP part of the equation, because they will use the local memory, they only use the system memory so they can create 16MB and 32MB versions of their 64MB card AND so game creators can create games that would like a 256MB video card and it will fake it... real coders target a platform, but I digress), we'd have REAL system IO. Instead was now have PCI Modems and Sound Cards (yeah, so important that we got them off the PCI bus...), PCI sound cards that use the PCI bus like vid. cards use the AGP bus to avoid putting RAM there, etc.
With RAM price this cheap, all my components should have some. Instead, All my devices want to graph system RAM (which is fine at 384MB), but that is sucking my my PCI bus from my SCSI card.
Next mobo will support the 64-bit, 66MHz PCI spec for my SCSI card, then I'll feel better.
The point is (as someone who also attends MIT, finishing my degree part-time), my entire MIT experience was the students that I was with. I found that many of my professors were useless, and had a terrible attendance record. I got my "education" from the web, books, and my fellow students. (ironically, the classes with good prof's were at 9AM and I slept through them anyways, oh well)
The material being online will fulfill two needs. It will make life easier on MIT students, knowing that all their information is easily available. It will avoid the stupid mess of courses lacking websites or having subpar ones, and as a side benefit (that they will trumpet for good media attention), they provide a public service.
MIT's school is solid and well rounded. However, outside the CS department (and some of the math department classes), the web sites are horrible, useless and outdated. Providing a real system for education for the students and those outside is great.
I don't want to question the intelligence of your peer group (and as Christian said, open 1000 flames), but it isn't the same. They may be just as bright, but they probably have to do work and stuff. MIT Students are famous for blowing off their work to help others. Having a school of insomniacs that are extremely intelligent and having nothing to do but school work and hang with friends is something you can't compare to working with intelligent people.
Even if your friends have the brain power, they are unlikely to have the time.
Slashdot is officially news for nerds. It is unofficially a Linux advocacy site. However, it also has GREAT information on technical issues. It is easier to read Slashdot then to follow 10 websites.
When a student, I could read Slashdot 5 times an hour to get every new story quickly. Now, some days I check twice, some times a few days before reading. Sometimes I'll post 10 times in a day, then not post for a money.
However, I also like keeping up on technical topics from Slashdot. However, it is a very anti-MS site. As an MCSE that got burned in the NT4 screw on the MCSE program, I have my own MS rants. However, he took a good opportunity to look good. When someone is reading Slashdot, they see that MS is being reasonable, and the kiddies on/. throwing tantrums marked up to a 5 shouting about Linux, and you get some good points.
He painted MS's blemishes in a good light. From someone whose server room is divided between NT4 Boxes and OpenBSD boxes, knowing the MS wants to fix it's mistakes is a good thing.
Most of the screaming on Slashdot is your high school and (to a lesser extent) college crowd. I remember futzing with Linux about 4 years ago, and I didn't mind that it took a few days to get everything to play nicely. At my last job, getting Linux configured to my liking took the better part of the day. Now that I sit on the other side of the employment divide, I realize the value of my employee's time. I want them enhancing the bottom line, not losing 4-6 hours getting sound to work.
This crowd talks about buying games. Most of the probably pirate them, or buy one copy and burn many copies. I found that as I have more money and less time, it isn't worth hacking anything. It is rarely worth the effort to get half assed solutions.
Professionally, at a job a while back we tried to mess with some open source stuff to adapt to our needs (including Slashcode!). Most of it was BEYOND subpar. The code being available was nice, but the documentation was lacking and the coding involved few abstraction layers so getting inside was a nightmare.
I have often found it easier to play with a few OSS packages, poke around when needed, but implement our own stuff. Having it available under the GPL is nice (to give back, we tossed a few of our building block pieces up, but never got around to polishing up the rest for release... maybe when we're not too busy) for learning, but most of the code out there is garbage.
I just dropped $130 to get MacOS X for my G4 Cube on my desk for playing around. It's nice, and I want to move over to it (to keep my development and application environment on one machine), but knowing Windows as well as I do (MCSE and NT Admin for the past 4 years, just moved back into the programming side of IT recently), it isn't worth it.
The problem is that people playing with Linux are mostly doing it as a hobby. Hobbyists don't want to spend money for the hobby, it's a labor of love. They learned Linux to learn Unix, and now find that they like some of the software and prefer to use it professionally.
However, despite all the GUI copying, KOffice is a joke compared to MS Office. When companies make an effort to develop something cool for Linux (VMWare, for starters, that was mentioned before), the OSS community immediately attempts a copy of it. That is somewhat self defeating. If you want software available right away on your platform, it would be nice to buy it instead of trying to copy everything.
Oh well, we'll just reimplement everything twice. Company 1 will make something useful and sell some copies. The OSS world will try to FUD them out of existance with a crappy substitute that will be good enough "real soon now".
Maybe the Open Source Advocates should start playing nice. Talk to Apple about GNUstep. Maybe if they toss some effort into it and OpenSTEP again, we'll have a real cross platform development environment. KDE/GNOME boys, you can provide your hooks.
Maybe we all write OpenSTEP applications to run on MacOS X, Linux/KDE, and Linux/GNOME, and when source is available, you can compile on *nix/KDE or *nix/GNOME... but maybe we'll all just flame Microsoft instead.
However, when a game purchase is a major portion of your income, $500 for Photoshop seems outrageously priced. But if you saw the salaries of your Graphic Designers... they get whatever they want.
Back when Internic was run by Network Solutions for the Commerce Department, there was opposition to their control. Alternic was proposed, with alternative TLDs if users (or ideally, ISPs) would switch their DNS over to support it.
This crashed and burned fabulously, but that may have been influenced by hack attempts on root servers, etc.
Still, this seems unlikely. People want.com. They don't want country codes, and they don't want the other minor ones.
When $100 was the registration fee and nobody was doing business online, people cared about cheap alternatives. Now that the costs of doing business on the Internet are huge, nobody is not going to shell out $70 (or $30/2 years with some of the cheaper services) while maintaining a presence.
Additionally, people want a good domain name or a generic one. As nobody is going to try to guess these random TLDs and assume that their ISP supports it, this names provide no value.
So I had already planned a vacation that left on March 24th, so no OS X on the trip. I picked up a copy towards the end of the vacation when we passed by a CompUSA. I got to install it yesterday... WOW.
First off, all the eye-candy is gorgeous. Now, I am running this on a reasonably current machine, Mac Cube G4-400 w/256MB RAM, but not a high end one (no dual processor, no 500+ MHz, so workstation with souped up video card). The Genie affect is neat, the quicktime movies playing in the dock are neat. Wow, doubly wow.
Now, I only got to play for a few hours, just poking through preferences and utilities. It's very cool. My work machine is still the Win2K box, but only until I configure this one. After this one is configured with programs, etc., then Win2K sits around for Visio only, and MacOS is the primary system.
BTW: my partner is a die-hard Unix geek. He was drooling. I don't think we'll buy another box for a Linux workstation. I mean, with xTools or XonX, there shouldn't be anything Open Source that won't run on it. I'm hoping that a Motif, GTK, and QT port to Quartz are all in the works so I can eventually get all my Unix stuff native.
Let's face it, if you are buying a new machine to run Linux (which I guess that most here don't do) as a workstation, why would you not buy a Mac for MacOS X. It will save you money, given the hours it takes me to have a Linux workstation approaching the usefulness and prettiness of the Mac.
Alex
Re:Didn't Steve Jobs Speak at MacWorld about....
on
Another Look At OS X
·
· Score: 2
Or Install Toast. My roommate has a Beige G3 and is looking into OS X "real soon". He's hoping that it is good enough to keep him on the Mac Platform. It's pulling me in as soon as I can afford one of the TiBooks.
However, Toast, which is apparently a large CD-Burning and CD-Imaging program, has an OS-X version?
It is an application that wasn't written yet because Apple want's it done right. Also, the Apple DVD and CD-burning software isn't run-of-the-mill stuff like their shitty Windows/Linux counterparts, it is an integrated system that allows drag-and-drop CD-burning, etc.
Expect the final versions to rock.
Just because the "built-in" software on the PC sucks doesn't mean that the Mac ones do.
The best thing I read was an attack on OS X making older machines obsolete. Their arguement? You can install Windows 95 on a machine going all the way back to around '92. Well, people attacking Apple neglect to remember that that is Microsoft's 5.5 year OLD operating system. Windows 2000 officially requires a Pentium processor, and that's only because it was supposed to ship in '97-'98. Realistically, you only run it on a P6 generation machine, or a few K6 generation machines if you are masochistic like I am.
Carbon will be the dominant environment for developing Macintosh applications for 3-5 years. That will run on OS 9.1 and OS X. In fact, I don't know if they've backported Carbon to OS 8.x, but I'd expect them to. The 8.1 supports back to the 68040 chip?
Either way, Apple has a much better history of supporting old machines. Yes, Linux does, but only because the modular Linux approach makes that easy. You can run the newer kernel on old machines, as the kernel is a small part of the system. Try to run modern programs on it.
The movie was utter garbage. Evven the extra long uncut version is crap. It doesn't cover the story, it unpleasant to watch, and requires reading the book to know what is going on.
I love Dune, read it a few times, read the first four books (was rereading them a few months ago to finish the series, but real life interfered), used to play on DuneMUSH II for hourse on end. But I hate the move.
The mini-series hwoever, was extremely entertaining AND true to the story. It captured the feel and flavor of Dune. I truly feel that you saw Dune. Sure you missed out on the Mentats, the true importants of Dr. Yueh's deception, etc., however, the essence of Dune (a plot within a plot) was there.
Sure their was silly plotting to feminize Irulan, but the story was there. I felt that this was Dune.
I have it sitting on tape just waitting to get a Dune-watching together.
Another reason that this series rocked? The original movie was painful, and sucked. It was unpleasant to watch, and if you hadn't read Dune, don't bother. This was fun even if you hadn't. My girlfriend watched it with me, and afterwords, she finally picked up my copy of Dune to read. This was after months of my prodding.
This mini-series was good, and probably got people to read Dune that hadn't yet.
I agree with you that Sen. Packwood's situation was attrocious. You can't be forced to testify against yourself, but we can claim your diary (your words to yourself) and use it against you?
We've essentially decided to give up any sense of privacy and protections against self-incrimination if it is written down. This is wrong.
Yeah, I particularly liked getting a letter from the local Congressman (Mass. liberal) explaining to me about how I can't get a tax cut on his tax-payer based correspondance note...
The real travesty is that it was provided for Congress to keep the locals informed when the House was representing the people and the political elites controlled the Senate and White House (officially, not just in reality).
It wasn't used much until computers... with targetted mail, Congress went nuts.
Slashdot saying something doesn't mean that it is true.
Without a doubt, they filter the letter-writing campaigns, because they are a waste of resources.
However, each congressman's budget includes "constituent work" budgets, which includes helping constituents deal with the Federal government, and responding to letters.
Why do they do that? Given the low turnout in congression elections (especially off-year), it is estimated that 5%-8% of the vote can be changed by these constituent actions.
As a congressman, I'm nuts to let my staff trash anything. While Slashdotters probably will not affect their vote by the thing, my father's poorly constructed prose will definitely affect his vote, etc.
The reason that they listen to letters is that they can't afford to alienate a vote.
Assume average American. Family of 4. One of them writes their congressman. If he gets a response, they and their spouse are likely to view that person more favorably. They are also more likely to defend them to their friends, etc. Assume each response generates 3 votes (0-6), if you can get 5%-8% of the voters to personally identify you because you responded/helped their friends and family, you are a shoe-in.
They would be insane to ignore e-mail. While your congressman doesn't personally read it, I can assure you that his staff has some people reading it, and letting him know what is going on.
The reason the RIAA freaked out about Napster, and the MPAA will be uncomfortable with this stuff, is the college market. After internal MP3 sharing (pre-Napster getting big, but people would run each other's MP3s over the network) was bringing my college fraternity (at MIT, so YMMV) old 10baseT network to a crawl. We couldn't rewire (the old buildings are a mess), so we brought switches to a at least have 100Megabit backbone and 10Megabit down.
Good thing two, because the following year (about 1.5 years ago), a freshman with a hobby for trading movies showed up. He managed to gather dozens of movies over our school provided T1, and he would share them across the house.
Forget that he would get them, it isn't worth it for most people to get them. But if one person acquires the movies (and keeps the archive on CDs to loan out) and makes them available to 40 friends... well...
Now make these 40 people all college kids with disposable income, and you've hit the target market. That's why they hate this. College kids are a big part of their market. High school kids are as well, and they are the most likely to get a DSL connection at home AND have lots of spare time.
Well, depends where you look. I mean, I wouldn't consider national college admissions as any indication of academics. It merely means that they continued school after high school.
If you look at gender statistics at the top schools, there is a slight male bias. In engineering, it is a pronounced male bias. That is the "top" academic achievements.
Indeed, if women are proportionally continueing onto more education, but the top schools are evenly divided or slightly male heavy, then it reinforces his belief that men are doing better.
Indeed, there are many reasons that would explain a higher attendance of females than males while still supporting his assertion.
For example, at lower income levels, a male is more likely to leave academics for exmployment. That will bias the results towards women.
Crime and incarceration rates are higher among males, and heavily pronounced among black males. This biases the results towards females.
Tight economy (we're still under 5% unemployment) makes this a better time to be in a job market. Assume a working class recent high school graduate, do they attend the local community college or take a job with decent wages. Community college can now wait.
The stereotypical MRS degree, girls off at school seeking spouses. This will result in non-academically oriented females continueing their education while their male counterparts are unlikely to do so. This also biases it towards female.
He was discussing academic performance along gender roles. You through out a meaningless trend (meaningless to this debate, not in general).
To put it in Slashdot terms, we're argueing about a portion of Linux's design, and someone points out that Linux is more stable than Win95... and claiming that this was debating.... Oh wait... that happens daily...
Perhaps you should look at a map... Start in Washington, go south of city limits, and you're in Virginia. Go several hundred miles and you should be able to reach North Carolina, depending on how many hundred.
Yes, become a game company. Sell an online service (monthly/hourly).
There has NEVER been another company that markets software to home users.
The entire PC phenomenon involved people being able to share data between home and work. The only things that people really bought were games. Most of their software either came on the PC when they took it home (remember when new computers shipped with Software? Before MS told them that they shouldn't do that), or they borrowed from the office. In a few cases, they buy a copy and give to their friends.
The entire PC Software market is based upon pirating software for the home market so that you establish a mindshare for the user and sell to the businesses.
You question, how to I make money developing software that I want to give away for free? My response, it can't be done.
Shareware CAN work, if you can get the businesses to pay.
Random tools USED to sell on the PC (disk doubler crap).
You need to sell a product or server that people will pay for that is greater than your costs.
Redhat would like that to be support contracts, but they don't have anything too useful. All they have is a dumb phone number/priority FTP server with the box. I haven't seen anything really good from them.
Also, if you write some really good GPL'd software that is useful, you have at most 3 months to ship it in volume to stores. Then RedHat includes it in their next point release of RedHat.
Sorry dude, find a real market. The Linux support market will exist, but the Linux software market won't.
Some other people compared the revenues of Redhat and VA to Microsoft and Oracle and said that it meant that licenses were all the money.
I would hazard a guess that if you combine ALL the revenue from supporting Microsoft products (all the consultants, integrators, etc), that much more money is made there than in the licenses. I would suggest that the same is true for Oracle.
The money is in the support contracts. Check with Sun.
Biggest "support" company? Compaq.
Whenever I want an NT Server, Compaq gets a call and ships me out my new Proliant. Why? Support. When I call their 800 number at 3 AM, I can get to a tech rep that will check their knowledge base, escalate my call quickly, or ship out a part quickly. There is no comparison.
You don't sell products, you sell solutions, and solutions are a server, often called support on Slashdot.
I have been an NT 4.0 MCSE for about 3 years at this point.
I have been working with NT Server 4.0 and Citrix WinFrame/MetaFrame for 3.5 years.
I have been a Citrix CCA for 1 year.
I have yet to see an install where the cost of labor (using a consultant) is not at LEAST 25% more than the costs of hardware and software, and with a credible company (I used to work for a fly-by-night MCSP, but it paid well), it is easily double.
A real NT4 network setup by experts or at least qualified morons used to set a company back $5000 or so for the network design, and actually setting up the PDC/BDC. Want Exchange? Gonna cost you another 1-2 thousand in costs.
It adds up.
MCSEs make less than Unix admins, but not by that much. Also, the MCSE is the only easy to use requirement for NT Admins, and the MCSE exams are a joke. The real NT admin knows the resource kit and registry reasonably well, including LOTS of command line tools and scripting. You can get an NT 4.0 MCSE (and I'd assume Win2K, but I don't use Win2K, so no need to recertify) without real knowledge.
Hence the low reputation of NT Admins.
It's not that NT Admins are dumb for being NT Admins. I mean, I would say that all things being equal, obtaining an MCSE makes you a bit more knowledgable (even with the damned test-prep books, I used the Resource Kit to learn), I mean, you'll learn something when studying. The problem is that things aren't equal, and unqualified people are becoming MCSEs.
MSDN, KB, etc., it find for a software developer, and it gets you mostly there. Realistically you need the giant super-duper MSDN package for developers, and I forget the price on it but it isn't that bad. If I want to support my application on Windows NT, I use this option. However, that doesn't help me get it on Linux. To get it on RedHat Linux, time to call RedHat. RedHat Linux IS an important platform these days.
Solaris is it's own mess. For starters, my admins make another $10K-$20K, and Ultra Sparc hardware is comparitively more expensive than Intel equivalents, sometimes significantly. I would suggest that my $2000-$3000 Linux server that includes mirrored IDE drivers, would require about $5000-$6000 for the equivalent Sun box. Getting Solaris happy on an Intel solution is a nightmare, and I haven't seen anyone selling Intel solutions for Solaris is a LONG time.
RedHat Linux may often be more expensive than Solaris/Win2K, it is then not the right option. If Solaris is the most cost effective, call up Sun's local rep. If Win2K is the solution, go to the computer store and pick it up or call your local MCSP.
Pick the most cost effective solution for your business and open up the checkbook. If you're in the Boston area, call Feratech.:)
Alex
Microsoft is DAMNED expensive. RedHat is damned expensive. Solaris is expensive. Not RedHat distributions aren't relevant (no corporate stability/marketshare/mindshare, sorry Caldera). Pick you poison, I chose OpenBSD... Although I still have an NT4 Server farm b/c I love Exchange.
Support is not a 900 number to call for technical support. Support is not for end users. End users buy a pretty little box with Wizards when they want "support". Support refers to business support.
What does this entail?
I'm developing a UNIX application aimed for Enterprise clients. I know that one of the platforms that I want to support includes Redhat Linux. I could count on my programmers spare time knowledge of Linux, but then it costs me a fortune in lost time if there is a problem. As a result, I pay them $20,000 for their special documentation which includes known bugs, etc., and a live number to talk to one of their kernel hackers to find out the problem.
Next scenario, I decide to migrate my expensive, but aging, HP-UX system to a modern, inexpensive system. I see that VA Linux will sell me a server for about the same price as Dell. VA Linux, however, will send out a technician (who as access to their knowledge base) and migrate my system for $6,000 including server. Otherwise, I risk losing my system for a few days while they iron out bugs.
Final scenario, I have 15 servers up and running, with a sysadmin that manages them including server side support for my IT guys. My IT team is writing software to improve our core business. However, they need DB support, etc. My sysadmin, however, spends 4 hours/day reading Slashdot/following bug news, applying patches, etc. That means that half his day is spent not aiding my core business.
As a solution? I could get another Sysadmin ($80k/year), or I pay Redhat $1000/server/year and they provide me with their enterprise system. Each of my servers are setup in their system, and all updates/bug fixes are automatically applied. I pay them $15,000, and I save the $40,000 of half a sysadmin.
That's support.
As a hobbyist, I don't mind spending 15 hours playing with my system until it's right. As a consultant, if I spend 15 hours on a problem, my company lost the ability to make $1500, and if a $500 or $1000 solution solves it? Well, we'll open the checkbook.
Sorry, I know that I've gotten slammed for this. What I MEANT was that I can run the *nix version, in fact a Linux binary version. That is actually a big deal. That means I can keep my development environment consistent.
While there shouldn't be a difference between Windows and *nix versions of Apache, if I'm developing/testing commercial software, I want to keep my environment consistent.
Wow, that's terrific, I sit here on my Win2K desktop, and I'm just thinking, "I love the OS for it's stability, if only I could natively run ALL those *nix applications!, then I'd have the perfect desktop".
</SARCASM>
In fairness, this is moderately useful, I could run Apache/PHP locally for development work on my standard desktop, but that's roughly where the usefullness ends. I mean, obviously the use for this isn't X-Windows applications, although with an X-Server running it would be.
While WINE is a useful project, this isn't. I give the people working on it props, it's a neat idea, and probably a fun hack, but doesn't seem terribly useful. I mean, while this is definitely different than the Cygwin project, I'm of the opinion that Cygwin provides 75%-85% of the benefits of a project like this, so the extra effort of this project isn't worth it to "the community". However, as it is worth it to the developers, good luck.
Actually, it blacks didn't count as 3/8s of a person. All other persons counted as 3/5s of a person, which meant any non-person person. Free Blacks were a full person, slave blacks were 3/5s. After the Civil War, the northern states freaked (hence reconstruction) when they realized that with all blacks counting as full persons, the south would become a dominate force in the House.
Actually, everyone doing clickthrough/sales commission advertisements now has standard text-based advertisements to include on your site. The advertisers have realized that text with links is FAR more useful than the banner ads, because people looking for information are more likely to read it than filter it out.
There was ONE line about the Simpsons. One line. The Slashdot summary had more text about the Simpsons than the article it was on.
The article covered their problems with market penetration and getting the advertisements in magazines.
There was as much on the Simpsons as the MUCH more interesting one-liner about how the magazines don't need to cater to advertisers as much because of the collapse of online advertisement as a competitor.
This was an absurd article, I'm disappointed in Slashdot for this one.
Good, maybe my programming staff will work...
on
KDE 2.1 Is Out
·
· Score: 2
Maybe my staff can actually do something more productive than grab the latest snapshot of the KDE Betas... I was impressed by them, but constant messing with the upgrades was aggravating.
I'm sure that Slashdoters will pay little attention to this, because for reasons that are beyond me the consensus here is pro-GNOME. Oh well, I think that KDE is so FAR ahead of GNOME it isn't even funny. The licensing issues were becoming a headache, so I'm glad Trolltech made them go away.
KOffice is the best shot they got at home office desktops. I wouldn't expect corporate adoption, but if the techies eventually have a reasonable environment and can share data with the rest of the corporation, there is no reason not to allow Linux desktops.
We need a method of free bandwidth. Ads and micropayments AREN'T the solution. Slashdot and Yahoo are among the few sites that make revenue of banner ads. Yahoo, because it's HUGE and if you want to hit everyone, they're the place to do it. Also with Yahoo, you can key off search words, and they are the most common search site. Besides, Yahoo collects money for reviewing sites and sponsered links.
Slashdot happens to have a GREAT target audiance. A good potion of it's viewers are people in decision-making positions regarding technology. I know that I spent over 10K on a business decision with a company that I found via Slashdot, I know that I'm not alone.
The real solution is cable TV style, but not networks of sites. Broadbands makes it possible.
In the dial-up world, phone lines were part of the infrastructure costs. People sitting online ate up limited resources. With DSL/Cable, there is no "cost" of being online, only bandwidth.
To make things more rediculous, bandwidth is asynchronous. This resulted in ISPs selling downstream bandwidth to consumers, and upstream bandwidth for web hosting, a hack system. The real solution is for ISPs to pay for content.
For example, Akamai has cacheing servers. Eventually, you host your content with Akamai or competitors, but instead of paying Akamai, Akamai pays you. Akamai charges the ISPs for the free "bandwidth" (if I connect from my ISP to the Internet, I use their expensive Internet connections, if I connect to a server sitting in their farm, I use no Internet bandwidth) because the ISPs pay less to Akamai than to their T3 providers.
In this model, I am paying not just for bandwidth, but for content/bandwidth. If my ISP has it local, I use no Internet bandwidth. If my ISP doesn't carry it, they "pay" for me to get it from their bandwidth.
With that model, ISPs try to maximize their free bandwidth (Cacheing) and minimize their expensive bandwidth (non-cacheing).
Think of it as proxy-serving on a mass scale.
For example, ISPs shouldn't be allowed to require a proxy server, they are distributing someone else's copyrighted work for a fee. They should be required to pay for the right to proxy, which would solve the situation.
Good content => more hits => more bandwidth => more money
Slashdot Mirror
AGP is a lousy system.
AGP is a cheap hack. It was created because it is slightly cheaper than 66 MHz 64-bit PCI (actually, quite a bit cheaper).
The 3dfx cards used AGP AS 66 MHz, 64-bit PCI (not using the AGP part).
We're NOT using AGP the way it was intended. The goal of AGP was to let you eliminate graphics memory by creating a direct link to system memory. Intel's goal it to make everything dependant on the primary system (CPU), and as a result, putting it to memory requirements is second best.
Intel wants to take the money AWAY from component makes and move it into CPUs, mobos, chipsets, and other core features that they dominate.
AGP isn't faster than 66MHZ 64-bit PCI, it just a less complicated approach.
The spec for server PCI has been on the Alphas for about 4 or 5 years now? That should have been the solution for new graphics cards. Instead of the ugly AGP hack (and REAL cards don't take advantage of the AGP part of the equation, because they will use the local memory, they only use the system memory so they can create 16MB and 32MB versions of their 64MB card AND so game creators can create games that would like a 256MB video card and it will fake it... real coders target a platform, but I digress), we'd have REAL system IO. Instead was now have PCI Modems and Sound Cards (yeah, so important that we got them off the PCI bus...), PCI sound cards that use the PCI bus like vid. cards use the AGP bus to avoid putting RAM there, etc.
With RAM price this cheap, all my components should have some. Instead, All my devices want to graph system RAM (which is fine at 384MB), but that is sucking my my PCI bus from my SCSI card.
Next mobo will support the 64-bit, 66MHz PCI spec for my SCSI card, then I'll feel better.
Alex
The point is (as someone who also attends MIT, finishing my degree part-time), my entire MIT experience was the students that I was with. I found that many of my professors were useless, and had a terrible attendance record. I got my "education" from the web, books, and my fellow students. (ironically, the classes with good prof's were at 9AM and I slept through them anyways, oh well)
The material being online will fulfill two needs. It will make life easier on MIT students, knowing that all their information is easily available. It will avoid the stupid mess of courses lacking websites or having subpar ones, and as a side benefit (that they will trumpet for good media attention), they provide a public service.
MIT's school is solid and well rounded. However, outside the CS department (and some of the math department classes), the web sites are horrible, useless and outdated. Providing a real system for education for the students and those outside is great.
I don't want to question the intelligence of your peer group (and as Christian said, open 1000 flames), but it isn't the same. They may be just as bright, but they probably have to do work and stuff. MIT Students are famous for blowing off their work to help others. Having a school of insomniacs that are extremely intelligent and having nothing to do but school work and hang with friends is something you can't compare to working with intelligent people.
Even if your friends have the brain power, they are unlikely to have the time.
Alex
Slashdot is officially news for nerds. It is unofficially a Linux advocacy site. However, it also has GREAT information on technical issues. It is easier to read Slashdot then to follow 10 websites.
/. throwing tantrums marked up to a 5 shouting about Linux, and you get some good points.
When a student, I could read Slashdot 5 times an hour to get every new story quickly. Now, some days I check twice, some times a few days before reading. Sometimes I'll post 10 times in a day, then not post for a money.
However, I also like keeping up on technical topics from Slashdot. However, it is a very anti-MS site. As an MCSE that got burned in the NT4 screw on the MCSE program, I have my own MS rants. However, he took a good opportunity to look good. When someone is reading Slashdot, they see that MS is being reasonable, and the kiddies on
He painted MS's blemishes in a good light. From someone whose server room is divided between NT4 Boxes and OpenBSD boxes, knowing the MS wants to fix it's mistakes is a good thing.
Alex
Because most don't have any money.
Most of the screaming on Slashdot is your high school and (to a lesser extent) college crowd. I remember futzing with Linux about 4 years ago, and I didn't mind that it took a few days to get everything to play nicely. At my last job, getting Linux configured to my liking took the better part of the day. Now that I sit on the other side of the employment divide, I realize the value of my employee's time. I want them enhancing the bottom line, not losing 4-6 hours getting sound to work.
This crowd talks about buying games. Most of the probably pirate them, or buy one copy and burn many copies. I found that as I have more money and less time, it isn't worth hacking anything. It is rarely worth the effort to get half assed solutions.
Professionally, at a job a while back we tried to mess with some open source stuff to adapt to our needs (including Slashcode!). Most of it was BEYOND subpar. The code being available was nice, but the documentation was lacking and the coding involved few abstraction layers so getting inside was a nightmare.
I have often found it easier to play with a few OSS packages, poke around when needed, but implement our own stuff. Having it available under the GPL is nice (to give back, we tossed a few of our building block pieces up, but never got around to polishing up the rest for release... maybe when we're not too busy) for learning, but most of the code out there is garbage.
I just dropped $130 to get MacOS X for my G4 Cube on my desk for playing around. It's nice, and I want to move over to it (to keep my development and application environment on one machine), but knowing Windows as well as I do (MCSE and NT Admin for the past 4 years, just moved back into the programming side of IT recently), it isn't worth it.
The problem is that people playing with Linux are mostly doing it as a hobby. Hobbyists don't want to spend money for the hobby, it's a labor of love. They learned Linux to learn Unix, and now find that they like some of the software and prefer to use it professionally.
However, despite all the GUI copying, KOffice is a joke compared to MS Office. When companies make an effort to develop something cool for Linux (VMWare, for starters, that was mentioned before), the OSS community immediately attempts a copy of it. That is somewhat self defeating. If you want software available right away on your platform, it would be nice to buy it instead of trying to copy everything.
Oh well, we'll just reimplement everything twice. Company 1 will make something useful and sell some copies. The OSS world will try to FUD them out of existance with a crappy substitute that will be good enough "real soon now".
Maybe the Open Source Advocates should start playing nice. Talk to Apple about GNUstep. Maybe if they toss some effort into it and OpenSTEP again, we'll have a real cross platform development environment. KDE/GNOME boys, you can provide your hooks.
Maybe we all write OpenSTEP applications to run on MacOS X, Linux/KDE, and Linux/GNOME, and when source is available, you can compile on *nix/KDE or *nix/GNOME... but maybe we'll all just flame Microsoft instead.
However, when a game purchase is a major portion of your income, $500 for Photoshop seems outrageously priced. But if you saw the salaries of your Graphic Designers... they get whatever they want.
Alex
Back when Internic was run by Network Solutions for the Commerce Department, there was opposition to their control. Alternic was proposed, with alternative TLDs if users (or ideally, ISPs) would switch their DNS over to support it.
.com. They don't want country codes, and they don't want the other minor ones.
This crashed and burned fabulously, but that may have been influenced by hack attempts on root servers, etc.
Still, this seems unlikely. People want
When $100 was the registration fee and nobody was doing business online, people cared about cheap alternatives. Now that the costs of doing business on the Internet are huge, nobody is not going to shell out $70 (or $30/2 years with some of the cheaper services) while maintaining a presence.
Additionally, people want a good domain name or a generic one. As nobody is going to try to guess these random TLDs and assume that their ISP supports it, this names provide no value.
Give it up, "The Man" 0wns DNS and won't let go.
Alex
So I had already planned a vacation that left on March 24th, so no OS X on the trip. I picked up a copy towards the end of the vacation when we passed by a CompUSA. I got to install it yesterday... WOW.
First off, all the eye-candy is gorgeous. Now, I am running this on a reasonably current machine, Mac Cube G4-400 w/256MB RAM, but not a high end one (no dual processor, no 500+ MHz, so workstation with souped up video card). The Genie affect is neat, the quicktime movies playing in the dock are neat. Wow, doubly wow.
Now, I only got to play for a few hours, just poking through preferences and utilities. It's very cool. My work machine is still the Win2K box, but only until I configure this one. After this one is configured with programs, etc., then Win2K sits around for Visio only, and MacOS is the primary system.
BTW: my partner is a die-hard Unix geek. He was drooling. I don't think we'll buy another box for a Linux workstation. I mean, with xTools or XonX, there shouldn't be anything Open Source that won't run on it. I'm hoping that a Motif, GTK, and QT port to Quartz are all in the works so I can eventually get all my Unix stuff native.
Let's face it, if you are buying a new machine to run Linux (which I guess that most here don't do) as a workstation, why would you not buy a Mac for MacOS X. It will save you money, given the hours it takes me to have a Linux workstation approaching the usefulness and prettiness of the Mac.
Alex
Or Install Toast. My roommate has a Beige G3 and is looking into OS X "real soon". He's hoping that it is good enough to keep him on the Mac Platform. It's pulling me in as soon as I can afford one of the TiBooks.
However, Toast, which is apparently a large CD-Burning and CD-Imaging program, has an OS-X version?
It is an application that wasn't written yet because Apple want's it done right. Also, the Apple DVD and CD-burning software isn't run-of-the-mill stuff like their shitty Windows/Linux counterparts, it is an integrated system that allows drag-and-drop CD-burning, etc.
Expect the final versions to rock.
Just because the "built-in" software on the PC sucks doesn't mean that the Mac ones do.
Alex
The best thing I read was an attack on OS X making older machines obsolete. Their arguement? You can install Windows 95 on a machine going all the way back to around '92. Well, people attacking Apple neglect to remember that that is Microsoft's 5.5 year OLD operating system. Windows 2000 officially requires a Pentium processor, and that's only because it was supposed to ship in '97-'98. Realistically, you only run it on a P6 generation machine, or a few K6 generation machines if you are masochistic like I am.
Carbon will be the dominant environment for developing Macintosh applications for 3-5 years. That will run on OS 9.1 and OS X. In fact, I don't know if they've backported Carbon to OS 8.x, but I'd expect them to. The 8.1 supports back to the 68040 chip?
Either way, Apple has a much better history of supporting old machines. Yes, Linux does, but only because the modular Linux approach makes that easy. You can run the newer kernel on old machines, as the kernel is a small part of the system. Try to run modern programs on it.
Alex
The movie was utter garbage. Evven the extra long uncut version is crap. It doesn't cover the story, it unpleasant to watch, and requires reading the book to know what is going on.
I love Dune, read it a few times, read the first four books (was rereading them a few months ago to finish the series, but real life interfered), used to play on DuneMUSH II for hourse on end. But I hate the move.
The mini-series hwoever, was extremely entertaining AND true to the story. It captured the feel and flavor of Dune. I truly feel that you saw Dune. Sure you missed out on the Mentats, the true importants of Dr. Yueh's deception, etc., however, the essence of Dune (a plot within a plot) was there.
Sure their was silly plotting to feminize Irulan, but the story was there. I felt that this was Dune.
I have it sitting on tape just waitting to get a Dune-watching together.
Another reason that this series rocked? The original movie was painful, and sucked. It was unpleasant to watch, and if you hadn't read Dune, don't bother. This was fun even if you hadn't. My girlfriend watched it with me, and afterwords, she finally picked up my copy of Dune to read. This was after months of my prodding.
This mini-series was good, and probably got people to read Dune that hadn't yet.
Alex
I agree with you that Sen. Packwood's situation was attrocious. You can't be forced to testify against yourself, but we can claim your diary (your words to yourself) and use it against you?
We've essentially decided to give up any sense of privacy and protections against self-incrimination if it is written down. This is wrong.
Alex
Yeah, I particularly liked getting a letter from the local Congressman (Mass. liberal) explaining to me about how I can't get a tax cut on his tax-payer based correspondance note...
The real travesty is that it was provided for Congress to keep the locals informed when the House was representing the people and the political elites controlled the Senate and White House (officially, not just in reality).
It wasn't used much until computers... with targetted mail, Congress went nuts.
Alex
Slashdot saying something doesn't mean that it is true.
Without a doubt, they filter the letter-writing campaigns, because they are a waste of resources.
However, each congressman's budget includes "constituent work" budgets, which includes helping constituents deal with the Federal government, and responding to letters.
Why do they do that? Given the low turnout in congression elections (especially off-year), it is estimated that 5%-8% of the vote can be changed by these constituent actions.
As a congressman, I'm nuts to let my staff trash anything. While Slashdotters probably will not affect their vote by the thing, my father's poorly constructed prose will definitely affect his vote, etc.
The reason that they listen to letters is that they can't afford to alienate a vote.
Assume average American. Family of 4. One of them writes their congressman. If he gets a response, they and their spouse are likely to view that person more favorably. They are also more likely to defend them to their friends, etc. Assume each response generates 3 votes (0-6), if you can get 5%-8% of the voters to personally identify you because you responded/helped their friends and family, you are a shoe-in.
They would be insane to ignore e-mail. While your congressman doesn't personally read it, I can assure you that his staff has some people reading it, and letting him know what is going on.
The reason the RIAA freaked out about Napster, and the MPAA will be uncomfortable with this stuff, is the college market. After internal MP3 sharing (pre-Napster getting big, but people would run each other's MP3s over the network) was bringing my college fraternity (at MIT, so YMMV) old 10baseT network to a crawl. We couldn't rewire (the old buildings are a mess), so we brought switches to a at least have 100Megabit backbone and 10Megabit down.
Good thing two, because the following year (about 1.5 years ago), a freshman with a hobby for trading movies showed up. He managed to gather dozens of movies over our school provided T1, and he would share them across the house.
Forget that he would get them, it isn't worth it for most people to get them. But if one person acquires the movies (and keeps the archive on CDs to loan out) and makes them available to 40 friends... well...
Now make these 40 people all college kids with disposable income, and you've hit the target market. That's why they hate this. College kids are a big part of their market. High school kids are as well, and they are the most likely to get a DSL connection at home AND have lots of spare time.
Alex
Well, depends where you look. I mean, I wouldn't consider national college admissions as any indication of academics. It merely means that they continued school after high school.
If you look at gender statistics at the top schools, there is a slight male bias. In engineering, it is a pronounced male bias. That is the "top" academic achievements.
Indeed, if women are proportionally continueing onto more education, but the top schools are evenly divided or slightly male heavy, then it reinforces his belief that men are doing better.
Indeed, there are many reasons that would explain a higher attendance of females than males while still supporting his assertion.
For example, at lower income levels, a male is more likely to leave academics for exmployment. That will bias the results towards women.
Crime and incarceration rates are higher among males, and heavily pronounced among black males. This biases the results towards females.
Tight economy (we're still under 5% unemployment) makes this a better time to be in a job market. Assume a working class recent high school graduate, do they attend the local community college or take a job with decent wages. Community college can now wait.
The stereotypical MRS degree, girls off at school seeking spouses. This will result in non-academically oriented females continueing their education while their male counterparts are unlikely to do so. This also biases it towards female.
He was discussing academic performance along gender roles. You through out a meaningless trend (meaningless to this debate, not in general).
To put it in Slashdot terms, we're argueing about a portion of Linux's design, and someone points out that Linux is more stable than Win95... and claiming that this was debating.... Oh wait... that happens daily...
Perhaps you should look at a map... Start in Washington, go south of city limits, and you're in Virginia. Go several hundred miles and you should be able to reach North Carolina, depending on how many hundred.
Yes, become a game company. Sell an online service (monthly/hourly).
There has NEVER been another company that markets software to home users.
The entire PC phenomenon involved people being able to share data between home and work. The only things that people really bought were games. Most of their software either came on the PC when they took it home (remember when new computers shipped with Software? Before MS told them that they shouldn't do that), or they borrowed from the office. In a few cases, they buy a copy and give to their friends.
The entire PC Software market is based upon pirating software for the home market so that you establish a mindshare for the user and sell to the businesses.
You question, how to I make money developing software that I want to give away for free? My response, it can't be done.
Shareware CAN work, if you can get the businesses to pay.
Random tools USED to sell on the PC (disk doubler crap).
You need to sell a product or server that people will pay for that is greater than your costs.
Redhat would like that to be support contracts, but they don't have anything too useful. All they have is a dumb phone number/priority FTP server with the box. I haven't seen anything really good from them.
Also, if you write some really good GPL'd software that is useful, you have at most 3 months to ship it in volume to stores. Then RedHat includes it in their next point release of RedHat.
Sorry dude, find a real market. The Linux support market will exist, but the Linux software market won't.
Some other people compared the revenues of Redhat and VA to Microsoft and Oracle and said that it meant that licenses were all the money.
I would hazard a guess that if you combine ALL the revenue from supporting Microsoft products (all the consultants, integrators, etc), that much more money is made there than in the licenses. I would suggest that the same is true for Oracle.
The money is in the support contracts. Check with Sun.
Biggest "support" company? Compaq.
Whenever I want an NT Server, Compaq gets a call and ships me out my new Proliant. Why? Support. When I call their 800 number at 3 AM, I can get to a tech rep that will check their knowledge base, escalate my call quickly, or ship out a part quickly. There is no comparison.
You don't sell products, you sell solutions, and solutions are a server, often called support on Slashdot.
Alex
Microsoft is NOT cheaper.
:)
I have been an NT 4.0 MCSE for about 3 years at this point.
I have been working with NT Server 4.0 and Citrix WinFrame/MetaFrame for 3.5 years.
I have been a Citrix CCA for 1 year.
I have yet to see an install where the cost of labor (using a consultant) is not at LEAST 25% more than the costs of hardware and software, and with a credible company (I used to work for a fly-by-night MCSP, but it paid well), it is easily double.
A real NT4 network setup by experts or at least qualified morons used to set a company back $5000 or so for the network design, and actually setting up the PDC/BDC. Want Exchange? Gonna cost you another 1-2 thousand in costs.
It adds up.
MCSEs make less than Unix admins, but not by that much. Also, the MCSE is the only easy to use requirement for NT Admins, and the MCSE exams are a joke. The real NT admin knows the resource kit and registry reasonably well, including LOTS of command line tools and scripting. You can get an NT 4.0 MCSE (and I'd assume Win2K, but I don't use Win2K, so no need to recertify) without real knowledge.
Hence the low reputation of NT Admins.
It's not that NT Admins are dumb for being NT Admins. I mean, I would say that all things being equal, obtaining an MCSE makes you a bit more knowledgable (even with the damned test-prep books, I used the Resource Kit to learn), I mean, you'll learn something when studying. The problem is that things aren't equal, and unqualified people are becoming MCSEs.
MSDN, KB, etc., it find for a software developer, and it gets you mostly there. Realistically you need the giant super-duper MSDN package for developers, and I forget the price on it but it isn't that bad. If I want to support my application on Windows NT, I use this option. However, that doesn't help me get it on Linux. To get it on RedHat Linux, time to call RedHat. RedHat Linux IS an important platform these days.
Solaris is it's own mess. For starters, my admins make another $10K-$20K, and Ultra Sparc hardware is comparitively more expensive than Intel equivalents, sometimes significantly. I would suggest that my $2000-$3000 Linux server that includes mirrored IDE drivers, would require about $5000-$6000 for the equivalent Sun box. Getting Solaris happy on an Intel solution is a nightmare, and I haven't seen anyone selling Intel solutions for Solaris is a LONG time.
RedHat Linux may often be more expensive than Solaris/Win2K, it is then not the right option. If Solaris is the most cost effective, call up Sun's local rep. If Win2K is the solution, go to the computer store and pick it up or call your local MCSP.
Pick the most cost effective solution for your business and open up the checkbook. If you're in the Boston area, call Feratech.
Alex
Microsoft is DAMNED expensive. RedHat is damned expensive. Solaris is expensive. Not RedHat distributions aren't relevant (no corporate stability/marketshare/mindshare, sorry Caldera). Pick you poison, I chose OpenBSD... Although I still have an NT4 Server farm b/c I love Exchange.
Support is not a 900 number to call for technical support. Support is not for end users. End users buy a pretty little box with Wizards when they want "support". Support refers to business support.
What does this entail?
I'm developing a UNIX application aimed for Enterprise clients. I know that one of the platforms that I want to support includes Redhat Linux. I could count on my programmers spare time knowledge of Linux, but then it costs me a fortune in lost time if there is a problem. As a result, I pay them $20,000 for their special documentation which includes known bugs, etc., and a live number to talk to one of their kernel hackers to find out the problem.
Next scenario, I decide to migrate my expensive, but aging, HP-UX system to a modern, inexpensive system. I see that VA Linux will sell me a server for about the same price as Dell. VA Linux, however, will send out a technician (who as access to their knowledge base) and migrate my system for $6,000 including server. Otherwise, I risk losing my system for a few days while they iron out bugs.
Final scenario, I have 15 servers up and running, with a sysadmin that manages them including server side support for my IT guys. My IT team is writing software to improve our core business. However, they need DB support, etc. My sysadmin, however, spends 4 hours/day reading Slashdot/following bug news, applying patches, etc. That means that half his day is spent not aiding my core business.
As a solution? I could get another Sysadmin ($80k/year), or I pay Redhat $1000/server/year and they provide me with their enterprise system. Each of my servers are setup in their system, and all updates/bug fixes are automatically applied. I pay them $15,000, and I save the $40,000 of half a sysadmin.
That's support.
As a hobbyist, I don't mind spending 15 hours playing with my system until it's right. As a consultant, if I spend 15 hours on a problem, my company lost the ability to make $1500, and if a $500 or $1000 solution solves it? Well, we'll open the checkbook.
Alex
Sorry, I know that I've gotten slammed for this. What I MEANT was that I can run the *nix version, in fact a Linux binary version. That is actually a big deal. That means I can keep my development environment consistent.
While there shouldn't be a difference between Windows and *nix versions of Apache, if I'm developing/testing commercial software, I want to keep my environment consistent.
Sorry for my unclear post.
Wow, that's terrific, I sit here on my Win2K desktop, and I'm just thinking, "I love the OS for it's stability, if only I could natively run ALL those *nix applications!, then I'd have the perfect desktop".
</SARCASM>
In fairness, this is moderately useful, I could run Apache/PHP locally for development work on my standard desktop, but that's roughly where the usefullness ends. I mean, obviously the use for this isn't X-Windows applications, although with an X-Server running it would be.
While WINE is a useful project, this isn't. I give the people working on it props, it's a neat idea, and probably a fun hack, but doesn't seem terribly useful. I mean, while this is definitely different than the Cygwin project, I'm of the opinion that Cygwin provides 75%-85% of the benefits of a project like this, so the extra effort of this project isn't worth it to "the community". However, as it is worth it to the developers, good luck.
Actually, it blacks didn't count as 3/8s of a person. All other persons counted as 3/5s of a person, which meant any non-person person. Free Blacks were a full person, slave blacks were 3/5s. After the Civil War, the northern states freaked (hence reconstruction) when they realized that with all blacks counting as full persons, the south would become a dominate force in the House.
Actually, everyone doing clickthrough/sales commission advertisements now has standard text-based advertisements to include on your site. The advertisers have realized that text with links is FAR more useful than the banner ads, because people looking for information are more likely to read it than filter it out.
There was ONE line about the Simpsons. One line. The Slashdot summary had more text about the Simpsons than the article it was on.
The article covered their problems with market penetration and getting the advertisements in magazines.
There was as much on the Simpsons as the MUCH more interesting one-liner about how the magazines don't need to cater to advertisers as much because of the collapse of online advertisement as a competitor.
This was an absurd article, I'm disappointed in Slashdot for this one.
Maybe my staff can actually do something more productive than grab the latest snapshot of the KDE Betas... I was impressed by them, but constant messing with the upgrades was aggravating.
I'm sure that Slashdoters will pay little attention to this, because for reasons that are beyond me the consensus here is pro-GNOME. Oh well, I think that KDE is so FAR ahead of GNOME it isn't even funny. The licensing issues were becoming a headache, so I'm glad Trolltech made them go away.
KOffice is the best shot they got at home office desktops. I wouldn't expect corporate adoption, but if the techies eventually have a reasonable environment and can share data with the rest of the corporation, there is no reason not to allow Linux desktops.
We need a method of free bandwidth. Ads and micropayments AREN'T the solution. Slashdot and Yahoo are among the few sites that make revenue of banner ads. Yahoo, because it's HUGE and if you want to hit everyone, they're the place to do it. Also with Yahoo, you can key off search words, and they are the most common search site. Besides, Yahoo collects money for reviewing sites and sponsered links.
Slashdot happens to have a GREAT target audiance. A good potion of it's viewers are people in decision-making positions regarding technology. I know that I spent over 10K on a business decision with a company that I found via Slashdot, I know that I'm not alone.
The real solution is cable TV style, but not networks of sites. Broadbands makes it possible.
In the dial-up world, phone lines were part of the infrastructure costs. People sitting online ate up limited resources. With DSL/Cable, there is no "cost" of being online, only bandwidth.
To make things more rediculous, bandwidth is asynchronous. This resulted in ISPs selling downstream bandwidth to consumers, and upstream bandwidth for web hosting, a hack system. The real solution is for ISPs to pay for content.
For example, Akamai has cacheing servers. Eventually, you host your content with Akamai or competitors, but instead of paying Akamai, Akamai pays you. Akamai charges the ISPs for the free "bandwidth" (if I connect from my ISP to the Internet, I use their expensive Internet connections, if I connect to a server sitting in their farm, I use no Internet bandwidth) because the ISPs pay less to Akamai than to their T3 providers.
In this model, I am paying not just for bandwidth, but for content/bandwidth. If my ISP has it local, I use no Internet bandwidth. If my ISP doesn't carry it, they "pay" for me to get it from their bandwidth.
With that model, ISPs try to maximize their free bandwidth (Cacheing) and minimize their expensive bandwidth (non-cacheing).
Think of it as proxy-serving on a mass scale.
For example, ISPs shouldn't be allowed to require a proxy server, they are distributing someone else's copyrighted work for a fee. They should be required to pay for the right to proxy, which would solve the situation.
Good content => more hits => more bandwidth => more money
Alex