Because this requires information from the DDC lines on the VGA connector. There isn't any standard interface for accessing this information as every vendor has a different way of sensing the information (I know how to do it with Matrox cards, but that's worthless for an ATI card.) The installer would have to fire up an X server to detect the video card and monitor assuming the DDC capability were there.
IMO, the installer should ASK the user to choose between an X install and a text install from the get-go. And then ask them the refresh rate for their display if they know what it is. Don't assume all of your users are complete freakin' morons. If you do then they will be the only ones who will use your product.
That "interactive startup" is just one more of the many annoying things about redhat. Why the **** does Redhat have to make a bunch of shell scripts so bloody complicated -- all they need to do is start and stop programs; the eye candy for the idiots who don't know what's going on anyway is purely useless ("fluff").
And this "kuydzu hardware detection utility"... how often does the hardware configuration of a machine change? What's the point in running this thing everytime the machine boots; don't you know when the hardware has been changed? Would you like it if windows ran a hardware discovery everytime the machine booted up?
"Security updates"... yes, alot of known problems have been fixed or patched. The real question is how many _new_ holes have they provided? In the last available version of Lorax (6.1 beta), "mv" didn't work right. I don't know if that's redhat's mistake or the fileutils people's mistake, but it's things like that that piss me off. (Apply a few seconds thought to things before you ship it.)
BUT, even with the growing spooge that is Redhat (and always has been), the installer is a nice thing.
Rule #1 applies to your kernel problem: Never use the supplied redhat kernel or source tree. Long ago, I stopped trying to figure out what those nuts did to the kernel they ship (why do they have to f*** with the kernel in the first place? If it's a valid, necessary change then send it in.)
Sounds alot like a holographic storage device. That would tend to make it rather hard to write data to it -- in any case, you'd never be able to write it as fast as you could read it.
Did they bother to tell anyone how they plan to populate the thing with data?
I remember an article about some "magnetic lens" technology a few years ago that boasted the same insane amounts of data storage -- it was R/W tho'.
Well, technically, BBS's pre-date CompuServ and other such information services. Of course, USENET would also qualify as prior art.
On the more obvious side, isn't it illegal (and surely immoral) to "hide" your patent for years and then "poof, you owe us money" after the patented idea/technology/whatever is everywhere? (That's what the whole "patent pending" deal is.)
15.5c/min! The ISP I used to work for paid $0.059/min with no trunk charge (i.e. if no one calls the 800# this month, we pay nothing at all.)
The bit about PRI's I can believe. They're a LD company, ISDN is not a LD technology:-( BTI had the same problem and the 800# still wasn't digitally signaled (it was forwarded from one switch to another to get it into a PRI.)
Why would you want a LD PRI? The odds are very low that a 64K connection could be made and only slightly better that a 56K connection could be made. (If it ain't end-to-end digital, it won't work.)
Funny, I signed up on-line. But yes, they ship "preconfigured" CDs -- they print a number on the label used to speed up the signup. And the signup runs you through the same thing as the online signup.
For the record, I use a netopia router.
As for ADSL, there's no reason for them to touch your computer at all (assuming you've already installed an ethernet card.) They need an ethernet MAC address to filter your line. No software is needed -- windows does DHCP at the push of a button (and reboot).
Because that would require an amount of clue on the part of the end user. If you've ever worked for an ISP, you'll agree the more you take away from idiot lusers and hide behind an icon, the better off you'll be.
Face it, 90% of the world doesn't know how to start "hyperterm" much less what to do once it's running. I'm all for the simple "BBS" style of account sign up, but the online pop. doesn't know how to do anything that doesn't involve IE or netscape.
"Make things idiot proof; we'll breed better idiots."
I used to work for an ISP. We played with setting up a netscape registration thing once, but abanonded it.
The simple reason for this being tied to 95/98 is on the dialup side. They ship you a file that installs a dialer to dial an 800# and then login with a special userid/password that sends you to a registration server.
We abandoned the stuff due to infrastructure overhead. We'd either have to have a different 800# for registration or setup a specific pool of addresses and a filter for that account. And as the current 800# could land you on several terminal servers (not all of them in the same place), it would have been a nightmare to keep straight.
In the end, the ISP set up a sep. 800# that's tied to a specific terminal server that sends you directly to the registration system -- the 800# doesn't cost anything but time (BTI's wierd that way) and there's one one TS to deal with. It is, of course, a windows based end-user system. Of course, the ISP has (always had) a web based signup as well as a faxed in "paper" signup.
I would have loved to ship a CD with a version of netscape for everything that's supported along with the 800# (no login required). The windows based crap can added the profile as necessary or let the user do it. Under other OSen, it's the users responsibility to get the thing dialed up, browser installed, and pointed to the right URL. (i.e. any URL)
Actually, the acoustic coupler was used during the call to the NSA. I doubt it was a modem, but still. I'll give 'em that one -- anything to throw some techno toys in there. Sure, there's alot of holloywood in there...
The bit about the "Russian codes" is completely true. If you were to build a device to cut through DES like water, then it's not going to help much with blowfish encrypted data.
Not that I doubt the NSA could recover a UNIX password in seconds, but your estimate of 30 years is grossly understated. Your average pentium (not pentium pro or II or III) can do about 20k crypt()s per second -- using ufc. Granted, this is not the most higly optimized DES setup one can find, but it's certainly the easiest to find.
There are 2^56 possible inputs (i.e. passwords).
(2^56/20000) / (60*60*24*365.25) == 114168.368377 YEARS
Assuming a highly optimized DES routine (and only DES processing time matters)...
(2^56) / (2000000/16) / (60*60*24*365.25) == 18266.938940 YEARS
Now, let's use DeepCrack...
(2^56) / (88000000000/16) / (60*60*24*365.25) == 0.415157 YEARS
The odds are good that you'll find it without trying every possible input.
Don't waste your time... RC5 on an alpha bites bigtime. DES, on the other hand, is a totally different story.
FWIW, sparc processors (ultra's esp.) are the same way. Makes me wonder why everyone is so quick to throw Sun enterprise servers out as web servers (esp. https:// servers.)
Now, if we can get Oracle to push out an Alpha/Linux database server, we'd be cookin' with nitro!
It was Digital who sent Linus an alpha some years ago to port linux to it. They even shipped to him on vacation in Australia of all places. (or so the story goes.)
Yeah, that thing kicks major ass. The error messages are complete, full, proper english sentences -- and they even make sense. (This assumes you turn on enough options.)
I was in awe of the Tru64 compiler. Of course, it's got a preprocessor macro expansion bug. (I guess I should report that:-))
There's nothing wrong with buying a compiler. If your business is making software (which you package and sell), buying a compiler is a very good investment.
GCC is designed to be portable more than optimal. I've never seen a commercial compiler that did not do a measurablly better job than a GNU compiler. This is because the compiler is designed specifically for that hardware and OS by people who get paid large sums of money to do things correctly, intellegently, and optimially.
If you're talking C++, then the GNU compiler is out of the question. If you write C++, buy a real [expletive] compiler.
Please remember your home dir is on an NFS array. move that file to/tmp (local disk) and try again. Also, run it more than once. A sample of 1 is not statistically valid.
Actually, there were quotas, just not OS enforced.
If you ate too much drive space for too long, you'd get a nasty-gram from an admin:-) I may still have one or two of those in my e-mail from college. (I was building Motif apps on one of them.)
What the h*** is wrong with you idiots?! Compaq very graciously put's a few dozen machines out so people can test their software on an Alpha instead of making them buy a $5k - $10k machine and associated OS -- and how much does Tru64 cost?
I applaude Compaq/Digital for doing this. As for the would be jack***es that will ruin this for everyone, do the world a favor and commit suicide.
Maybe some of you will remember axp.pa.dec.com. Years ago, digital had two Alphas set aside for people to test their software -- completely unrestricted (those testdrive boxes are very closed) -- with accounts that didn't expire in 30 days. As long as you kept using it, digital would leave the account active.
Would you go out and buy a top-of-the-line VAX if you had no experience with one -- and zero experience with VMS?
To answer your question... NO. Compaq will simply put the machine(s) back in their internal testing group and never again offer free access to their hardware to the general public.
For the record, I do own an Alpha Server. It runs Linux (or tries to:-)) I'd like to have a copy of Digital UNIX (Tru64) But that's a large chunk of cash for a "toy".
"linux" is the kernel; Redhat, Debian, Slackware, SLS (*grin*), SuSe, etc. are OSen.
NT "out of the box" (read: straight off the CD) is far more problematic than most Linux distributions "out of the box". How many service packs and/or hotfixes are required to keep NT 4.0 from walking off a cliff? [Redhat is a bad example, but I'll use it anyway.] How many updates are required to keep Redhat 4.2 from jogging into on-coming traffic? In both cases, you will need to turn a few things on or off depending on what you selected during installation. (And in the NT 4.0 case, you need to install the 70M IE4 to get it near usable -- it shipped with IE3 which cannot be used to access even Microsofts download section(s). I find that damned annoying.)
Kernel to Kernel, linux and NT are too close to call. Just look at how often kernel related defects for both systems turn up. Which is more secure? Neither. Both systems can be compromised -- it's generally easier on a linux system due to the ease of (nearly) replicating the system and the availablity of code to thumb through. (It's hard to break into a black-box.)
Givin a choice, I'll take any UNIX over Windows. I like having a command line; I hate having magic hidden behind GUI buttons; And I _like_ being able to "telnet" into my UNIX server that has no video device at all.
"I don't care if a pair of gerbals could break into it; I'm gonna use linux."
The 300 msgs per month you're getting at hotmail aren't all from NIC records. Where spammers harvest email addresses still baffles me -- addresses that have never been published _anywhere_ have gotten SPAM. Maybe it was a lucky guess? Maybe Bellsouth sold their subscriber list?
I agree with you, however, in that the agreement doesn't do a damn thing to prevent spammers from grabbing email addresses if NSI never actually stands up and enforces it. Personally, I rather they not publish email addresses at all. The only reason they're there is for them to contact you.
This is yet another "good thing" that's been perverted by a handful of morons out to make three cents more than anyone else without lifting a finger.
This is certainly one of my pet peeves... IDIOT firewall admins who turn off ICMP entirely, or worse, turn off everything but ICMP echo. ICMP exists for a reason. Sure, there are some parts that can be turned off without a problem, but you need to understand what the f*** you're doing -- which is at least 105% of the problem.
As for that BS about routers using "ping"... no they don't -- or more accurately, none that are worth their weight in twinkies. There are much better ways to judge distance -- oh, say, like the TTL in any IP packet. (Note: bind does this already.) Additionally, if you knew anything at all about ICMP, you'd know there is no (zero, none!) transmittion assurance for ICMP traffic. Nothing is going to alert you an ICMP message never got to it's dest. nor will anything ever retransmit an ICMP message. (RFC) Rule #1: NEVER SEND AN ICMP MESSAGE ABOUT AN ICMP MESSAGE.
For the record, the parts that can be turned off without breaking the network stack are:
First off, finding any site(s) without banner ads of some sort is nearly impossible these days. Internet connectivity isn't free. Web hosting isn't free either -- and don't even get me started on those "free" sites that do nothing but litter one's desktop with ads...
/.is (or was) a hobby site. As such, do you honestly expect people to fork out their (hard earned?) cash for redundant, "it will never go down", high availablity hardware? If you do, then you should seek professional help.
Let us do some math in reference to banner ads and "hobby sites"...
Let's make a few assumptions (alright, ALOT of assumptions.) Say your hobby site is fairly popular getting 100k hits per day with an average transfer of 8k per hit. <math> 100000 * 8 / (60 * 60 * 24) == 9.26K/s</math> That's assuming the traffic is perfectly evenly distributed -- which we all know isn't true.
To continuously transmit 9.26K/s would require a 128k connection. A standard analog modem tops out in the 3K/s range. A single ISDN B channel tops out at 7.5K/s. A dual channel ISDN (2B) or 128k frame connection tops out at 15.2K/s. Let's assume you go with ISDN (it's generally the lower cost.) SO, you'd need an ISDN router (the first person to say "ISDN Modem" gets shot) at a cost of 700$US, an ISDN phone line at a cost of 100$US/month and 200$US installation, and an ISP at a cost of 150$US/month plus 200$US installation (what a rip-off.) That's 1100$US to start plus 250$US per month for your "hobby"... and you haven't even bought a machine to be your web server.
Seeing as no one is being charged for access to/., no one has a leg to stand on if/when it fails. Face it, machines break. I think the/.'ers put more than enough of their time and money into making this geek haven. If you cannot get your/. fix, then that should be taken as an omen to get (back) to work.
Slashdot Mirror
(I thought of that too...)
Because this requires information from the DDC lines on the VGA connector. There isn't any standard interface for accessing this information as every vendor has a different way of sensing the information (I know how to do it with Matrox cards, but that's worthless for an ATI card.) The installer would have to fire up an X server to detect the video card and monitor assuming the DDC capability were there.
IMO, the installer should ASK the user to choose between an X install and a text install from the get-go. And then ask them the refresh rate for their display if they know what it is. Don't assume all of your users are complete freakin' morons. If you do then they will be the only ones who will use your product.
That "interactive startup" is just one more of the many annoying things about redhat. Why the **** does Redhat have to make a bunch of shell scripts so bloody complicated -- all they need to do is start and stop programs; the eye candy for the idiots who don't know what's going on anyway is purely useless ("fluff").
And this "kuydzu hardware detection utility"... how often does the hardware configuration of a machine change? What's the point in running this thing everytime the machine boots; don't you know when the hardware has been changed? Would you like it if windows ran a hardware discovery everytime the machine booted up?
"Security updates"... yes, alot of known problems have been fixed or patched. The real question is how many _new_ holes have they provided? In the last available version of Lorax (6.1 beta), "mv" didn't work right. I don't know if that's redhat's mistake or the fileutils people's mistake, but it's things like that that piss me off. (Apply a few seconds thought to things before you ship it.)
BUT, even with the growing spooge that is Redhat (and always has been), the installer is a nice thing.
Rule #1 applies to your kernel problem: Never use the supplied redhat kernel or source tree. Long ago, I stopped trying to figure out what those nuts did to the kernel they ship (why do they have to f*** with the kernel in the first place? If it's a valid, necessary change then send it in.)
Sounds alot like a holographic storage device. That would tend to make it rather hard to write data to it -- in any case, you'd never be able to write it as fast as you could read it.
Did they bother to tell anyone how they plan to populate the thing with data?
I remember an article about some "magnetic lens" technology a few years ago that boasted the same insane amounts of data storage -- it was R/W tho'.
Well, technically, BBS's pre-date CompuServ and other such information services. Of course, USENET would also qualify as prior art.
On the more obvious side, isn't it illegal (and surely immoral) to "hide" your patent for years and then "poof, you owe us money" after the patented idea/technology/whatever is everywhere? (That's what the whole "patent pending" deal is.)
Personally, I'd say revoke or nullify every patent they've ever filed and never let them file another one, EVER.
That should be enough of a "holy shit" to prevent any idiot from filing lame patents -- esp. firms with lots of patents.
15.5c/min! The ISP I used to work for paid $0.059/min with no trunk charge (i.e. if no one calls the 800# this month, we pay nothing at all.)
:-( BTI had the same problem and the 800# still wasn't digitally signaled (it was forwarded from one switch to another to get it into a PRI.)
The bit about PRI's I can believe. They're a LD company, ISDN is not a LD technology
Why would you want a LD PRI? The odds are very low that a 64K connection could be made and only slightly better that a 56K connection could be made. (If it ain't end-to-end digital, it won't work.)
Funny, I signed up on-line. But yes, they ship "preconfigured" CDs -- they print a number on the label used to speed up the signup. And the signup runs you through the same thing as the online signup.
For the record, I use a netopia router.
As for ADSL, there's no reason for them to touch your computer at all (assuming you've already installed an ethernet card.) They need an ethernet MAC address to filter your line. No software is needed -- windows does DHCP at the push of a button (and reboot).
Because that would require an amount of clue on the part of the end user. If you've ever worked for an ISP, you'll agree the more you take away from idiot lusers and hide behind an icon, the better off you'll be.
Face it, 90% of the world doesn't know how to start "hyperterm" much less what to do once it's running. I'm all for the simple "BBS" style of account sign up, but the online pop. doesn't know how to do anything that doesn't involve IE or netscape.
"Make things idiot proof; we'll breed better idiots."
I used to work for an ISP. We played with setting up a netscape registration thing once, but abanonded it.
The simple reason for this being tied to 95/98 is on the dialup side. They ship you a file that installs a dialer to dial an 800# and then login with a special userid/password that sends you to a registration server.
We abandoned the stuff due to infrastructure overhead. We'd either have to have a different 800# for registration or setup a specific pool of addresses and a filter for that account. And as the current 800# could land you on several terminal servers (not all of them in the same place), it would have been a nightmare to keep straight.
In the end, the ISP set up a sep. 800# that's tied to a specific terminal server that sends you directly to the registration system -- the 800# doesn't cost anything but time (BTI's wierd that way) and there's one one TS to deal with. It is, of course, a windows based end-user system. Of course, the ISP has (always had) a web based signup as well as a faxed in "paper" signup.
I would have loved to ship a CD with a version of netscape for everything that's supported along with the 800# (no login required). The windows based crap can added the profile as necessary or let the user do it. Under other OSen, it's the users responsibility to get the thing dialed up, browser installed, and pointed to the right URL.
(i.e. any URL)
Actually, the acoustic coupler was used during the call to the NSA. I doubt it was a modem, but still. I'll give 'em that one -- anything to throw some techno toys in there. Sure, there's alot of holloywood in there...
The bit about the "Russian codes" is completely true. If you were to build a device to cut through DES like water, then it's not going to help much with blowfish encrypted data.
Look at DeepCrack...
Not that I doubt the NSA could recover a UNIX password in seconds, but your estimate of 30 years is grossly understated. Your average pentium (not pentium pro or II or III) can do about 20k crypt()s per second -- using ufc. Granted, this is not the most higly optimized DES setup one can find, but it's certainly the easiest to find.
There are 2^56 possible inputs (i.e. passwords).
(2^56/20000) / (60*60*24*365.25) == 114168.368377 YEARS
Assuming a highly optimized DES routine (and only DES processing time matters)...
(2^56) / (2000000/16) / (60*60*24*365.25) == 18266.938940 YEARS
Now, let's use DeepCrack...
(2^56) / (88000000000/16) / (60*60*24*365.25) == 0.415157 YEARS
The odds are good that you'll find it without trying every possible input.
Don't waste your time... RC5 on an alpha bites bigtime. DES, on the other hand, is a totally different story.
FWIW, sparc processors (ultra's esp.) are the same way. Makes me wonder why everyone is so quick to throw Sun enterprise servers out as web servers (esp. https:// servers.)
Now, if we can get Oracle to push out an Alpha/Linux database server, we'd be cookin' with nitro!
s/Compaq/Digital/
It was Digital who sent Linus an alpha some years ago to port linux to it. They even shipped to him on vacation in Australia of all places. (or so the story goes.)
Yeah, that thing kicks major ass. The error messages are complete, full, proper english sentences -- and they even make sense. (This assumes you turn on enough options.)
:-))
I was in awe of the Tru64 compiler. Of course, it's got a preprocessor macro expansion bug. (I guess I should report that
There's nothing wrong with buying a compiler. If your business is making software (which you package and sell), buying a compiler is a very good investment.
GCC is designed to be portable more than optimal. I've never seen a commercial compiler that did not do a measurablly better job than a GNU compiler. This is because the compiler is designed specifically for that hardware and OS by people who get paid large sums of money to do things correctly, intellegently, and optimially.
If you're talking C++, then the GNU compiler is out of the question. If you write C++, buy a real [expletive] compiler.
Please remember your home dir is on an NFS array. move that file to /tmp (local disk) and try again. Also, run it more than once. A sample of 1 is not statistically valid.
Actually, there were quotas, just not OS enforced.
:-) I may still have one or two of those in my e-mail from college. (I was building Motif apps on one of them.)
If you ate too much drive space for too long, you'd get a nasty-gram from an admin
What the h*** is wrong with you idiots?! Compaq very graciously put's a few dozen machines out so people can test their software on an Alpha instead of making them buy a $5k - $10k machine and associated OS -- and how much does Tru64 cost?
:-)) I'd like to have a copy of Digital UNIX (Tru64) But that's a large chunk of cash for a "toy".
I applaude Compaq/Digital for doing this. As for the would be jack***es that will ruin this for everyone, do the world a favor and commit suicide.
Maybe some of you will remember axp.pa.dec.com. Years ago, digital had two Alphas set aside for people to test their software -- completely unrestricted (those testdrive boxes are very closed) -- with accounts that didn't expire in 30 days. As long as you kept using it, digital would leave the account active.
Would you go out and buy a top-of-the-line VAX if you had no experience with one -- and zero experience with VMS?
To answer your question... NO. Compaq will simply put the machine(s) back in their internal testing group and never again offer free access to their hardware to the general public.
For the record, I do own an Alpha Server. It runs Linux (or tries to
note to worldnetdaily... it's "encoding" NOT "encryption". What part of writing 1's and 0's as black bars is "encryption"?
I suppose melting ice into water is alchemy?
"carry around a barcode reader"? How many people carry around Palm Pilots? (or any IR capable computer-ish device.)
It doesn't take minutes, it takes seconds. It's just like reading a foreign language, assembly code, or even machine byte code.
"linux" is the kernel; Redhat, Debian, Slackware, SLS (*grin*), SuSe, etc. are OSen.
NT "out of the box" (read: straight off the CD) is far more problematic than most Linux distributions "out of the box". How many service packs and/or hotfixes are required to keep NT 4.0 from walking off a cliff? [Redhat is a bad example, but I'll use it anyway.] How many updates are required to keep Redhat 4.2 from jogging into on-coming traffic? In both cases, you will need to turn a few things on or off depending on what you selected during installation. (And in the NT 4.0 case, you need to install the 70M IE4 to get it near usable -- it shipped with IE3 which cannot be used to access even Microsofts download section(s). I find that damned annoying.)
Kernel to Kernel, linux and NT are too close to call. Just look at how often kernel related defects for both systems turn up. Which is more secure? Neither. Both systems can be compromised -- it's generally easier on a linux system due to the ease of (nearly) replicating the system and the availablity of code to thumb through. (It's hard to break into a black-box.)
Givin a choice, I'll take any UNIX over Windows. I like having a command line; I hate having magic hidden behind GUI buttons; And I _like_ being able to "telnet" into my UNIX server that has no video device at all.
"I don't care if a pair of gerbals could break into it; I'm gonna use linux."
Does no one else remember the Rats of Nim? *grin*
The 300 msgs per month you're getting at hotmail aren't all from NIC records. Where spammers harvest email addresses still baffles me -- addresses that have never been published _anywhere_ have gotten SPAM. Maybe it was a lucky guess? Maybe Bellsouth sold their subscriber list?
I agree with you, however, in that the agreement doesn't do a damn thing to prevent spammers from grabbing email addresses if NSI never actually stands up and enforces it. Personally, I rather they not publish email addresses at all. The only reason they're there is for them to contact you.
This is yet another "good thing" that's been perverted by a handful of morons out to make three cents more than anyone else without lifting a finger.
This is certainly one of my pet peeves... IDIOT firewall admins who turn off ICMP entirely, or worse, turn off everything but ICMP echo. ICMP exists for a reason. Sure, there are some parts that can be turned off without a problem, but you need to understand what the f*** you're doing -- which is at least 105% of the problem.
/* Echo Reply */ /* Echo Request */ /* Information Request */ /* Information Reply */ /* Address Mask Request */ /* Address Mask Reply */
As for that BS about routers using "ping"... no they don't -- or more accurately, none that are worth their weight in twinkies. There are much better ways to judge distance -- oh, say, like the TTL in any IP packet. (Note: bind does this already.) Additionally, if you knew anything at all about ICMP, you'd know there is no (zero, none!) transmittion assurance for ICMP traffic. Nothing is going to alert you an ICMP message never got to it's dest. nor will anything ever retransmit an ICMP message. (RFC) Rule #1: NEVER SEND AN ICMP MESSAGE ABOUT AN ICMP MESSAGE.
For the record, the parts that can be turned off without breaking the network stack are:
ICMP_ECHOREPLY 0
ICMP_ECHO 8
ICMP_INFO_REQUEST 15
ICMP_INFO_REPLY 16
ICMP_ADDRESS 17
ICMP_ADDRESSREPLY 18
Note: Turning off address mask info will break HP open spew.
First off, finding any site(s) without banner ads of some sort is nearly impossible these days. Internet connectivity isn't free. Web hosting isn't free either -- and don't even get me started on those "free" sites that do nothing but litter one's desktop with ads...
/. is (or was) a hobby site. As such, do you honestly expect people to fork out their (hard earned?) cash for redundant, "it will never go down", high availablity hardware? If you do, then you should seek professional help.
/., no one has a leg to stand on if/when it fails. Face it, machines break. I think the /.'ers put more than enough of their time and money into making this geek haven. If you cannot get your /. fix, then that should be taken as an omen to get (back) to work.
Let us do some math in reference to banner ads and "hobby sites"...
Let's make a few assumptions (alright, ALOT of assumptions.) Say your hobby site is fairly popular getting 100k hits per day with an average transfer of 8k per hit. <math> 100000 * 8 / (60 * 60 * 24) == 9.26K/s</math> That's assuming the traffic is perfectly evenly distributed -- which we all know isn't true.
To continuously transmit 9.26K/s would require a 128k connection. A standard analog modem tops out in the 3K/s range. A single ISDN B channel tops out at 7.5K/s. A dual channel ISDN (2B) or 128k frame connection tops out at 15.2K/s. Let's assume you go with ISDN (it's generally the lower cost.)
SO, you'd need an ISDN router (the first person to say "ISDN Modem" gets shot) at a cost of 700$US, an ISDN phone line at a cost of 100$US/month and 200$US installation, and an ISP at a cost of 150$US/month plus 200$US installation (what a rip-off.) That's 1100$US to start plus 250$US per month for your "hobby"... and you haven't even bought a machine to be your web server.
Seeing as no one is being charged for access to
As always, Your Milage MUST Vary!