One of our mail servers was blocked by Comcast for a while. I'd a nice chat with one of their "technical supervisors" as a result.
I suggested that they offer a feedback solution like AOL's. He wanted to, but Comcast wasn't willing to spend the money on it. I then suggested that he let clients specify their own filtering rules. He wanted to do this too, but Comcast wasn't willing to spend the money on it.
We went around a few times this was before I realized that there was a dishonesty here. This isn't about protecting clients' mail boxes, despite what they're saying at Comcast. This is about cutting their costs. Spam requires handling. So they're just blocking anything that looks like spam to keep the mail server load lower than otherwise.
I suggested that some clients wanted to do their own filtering. That's why my servers will forward spam: we do have client specific filtering, and some clients want little or none on the server.
In other words, those Comcast clients that had expressedly told us to use little/no server-side filtering on their email were not being properly served by Comcast because Comcast is forcing its filtering upon those Comcast clients against their will. Worse, it could - in fact, does - cause these Comcast clients to miss messages they want to receive.
The answer to this was that anyone that was dependent upon email that way shouldn't be using a "consumer grade service". He implied that Comcast's business service didn't have this problem (although he wasn't sure).
So not only is Comcast looking to save money by cutting the load on their mail servers, they're downgrading their "consumer grade service" w/o cutting the price of this service. More, they are trying to push customers that complain into paying more.
You can explain how if it takes you twice as long to make sure it is easy to maintain in the future and will save money in the long run until you blue in the face. But what it comes down to is if your inital cost of development is less then the other guy.
This is why I love clients that have been previously burned by some code & cash developer [team]. They understand what the consequences of "low quality" mean, and are willing to pay to avoid it.
The more educated the client, the higher the demand for quality. It's just a shame that so many need to be burned to be educated.
Let me explain this as clearly as possible: business is about money.
You're young (or have been incredibly lucky in your experience).
[Or perhaps I've merely been unlucky.]
You speak of corporate strategy as if it is some mystical force beyond the ken of mere technologists. Yet you also accuse technologists of arrogance, which is nicely ironic.
Having some small experience along the technology/management divide, I can tell you that there are two issues you've missed. First, you missed the impact of the personal strategy of the manager(s) involved. I've seen projects impacted because of empire building, for example, or even just simple self-preservation. At a planning meeting for the installation of some software at a regional office, the manager of that office calmly explained that he'd support us completely until any capability of the new software threatened his job. At that point, he'd do everything he could to cause the project to fail.
Nicely aligned with corporate stragety there, eh?
I recall on another project having to waste several hours a week over a period of months defending a particular hardware choice. The choice we made was the obviously correct one, in that the resulting product was cheaper and more capable. But it did mean not exploiting an internal product.
We did consider the balance, but net it was still more profitable for the company to not use the internal product in this case. Yet meeting after meeting after meeting required that we explain this to the various "internal salespeople" for that product.
I eventually left that company over issues like this. The VP for whom I was working at the time tried to either convince me to stay or salve his own ego (I've never known which) by telling me a story. Back when he was an engineer, he'd had a similar choice: use an internal chip set or one from (I think) TI. The internal set cost on the order of a dollar per set; that one from outside cost on the order of pennies.
[Note: It's been 20 years; I may not recall those numbers accurately.]
Being the good company man, he chose internal.
I asked if the product had ever gone to market. It hadn't. It was too expensive to compete.
He told this to me, by the way, with a straight face. He actually [appeared to] still believe that he'd made the right choice.
Good corporate strategy there, eh?
I don't really mean to belittle that VP. He really was just following the strategy laid out for him. And, outside of issues like this, I recall him being a good manager. It was the strategy itself that was flawed in this case. Followed, it lost money and business for the company.
Of course, I'm assuming that the expression of this strategy was reaching us accurately. That need not have been the case. Perhaps some higher level manager (acting out of a personal strategy) warped it to serve his own interests. I'll never know.
Please don't misunderstand this to be a claim that technologists are never at fault for failures. I've other beefs with many of them. Working in the financial software world for a number of years, I was surprised at the number of software people that were completely disinterested in learning about the instruments whose behaviors they were modeling. If nothing else, it was *math*. What software engineer or programmer wouldn't enjoy the math of financial modeling?!?
I remain convinced that, at least in many cases, it was a form of arrogance that kept people disinterested.
Nevertheless, I felt I could not let your commentary slide by. Your idealistic view of a business as being honestly about profit is incomplete at best.
Doesn't code with lots of syntax errors turn out to have more bugs once it finally compiles? Why wouldn't the same be true for human language?
By all rights, engineers - people trained to work w/in system constraints or else - should be excellent writers. Rules of grammar and the like are conceptually no different than those of any engineering discipline. Prose is just another system.
Why this is so rarely the case I've never understood.
"Difference is the US doesn't deny this to its citizens."
Do you think that difference is worth a hill of beans?
I do. So does the current administration. That's why they're now looking to prosecute reporters that "hurt national security" by telling Americans the truth about our government.
By contrast, the typical American supports human rights for both herself and others.
Unless the others include homosexuals that want to marry or pregnant women that don't want to be. Opposition to rights for these people can apparently be used to entice people to vote who'd otherwise not care enough about America to do so.
Anyone would be reviled in the US in public for forming an organization against civil rights for (for example) Blacks or the Irish. But homosexuals and pregnant women are apparently fair game. They're the real others today.
This is the power of "language survey" courses. As much as they provide an overview of the various languages, the real strength is that they teach the differences. How is LISP different from Java different from SML? What difference do those differences make, and how does it alter one's approach to problem solving.
Another advantage is that it provides access to a set of "problem solving" approaches that might not be "natural" in a given language but which might still be the best way to think about a problem.
I agree with those writing to skip the IDE in an introductory course. It's a distraction. Further, the advantages an IDE provides only truly appear with more complex software - software not likely to be written in an introductory course. Therefore, using the IDE that early (1) takes time away from the primary goal of the class and (2) forces the student to use a tool that provides little advantage.
Better to teach an IDE concurrent with more advanced programming projects that can really benefit from such a tool.
You cite the cost of that, and that is a valid concern. However, consider that no one proponent of this program will face the complexity. Using your example, each of the five networks will bill Google individually. There's some billing cost, but it should scale well: only paying "clients" are billed.
The complexity you cite will be faced instead by the content providers. In your example, it is Google that must track costs and issue payments to five different connectivity providers.
And Google already sees this as a Bad Idea. So while I agree that you're presenting a real cost, it's not a cost about which the proponents of this plan are going to be concerned.
However, my email client, my video player, and my web browser still run with the full privilege of my user account, when something less would be sufficient.
This is important, as many forms of malware (including that needed to build a 'bot) can be implemented w/o the requirement of root/superuser access. While the OS protecting itself is a Good Thing, this doesn't do anything to protect the computer itself against abuse (or to protect the Internet against abuse of this computer).
This is a fact too often missed during these discussions. And it's why we do need "least privilege", sandboxing, etc. for applications which execute untrusted content.
Further, comparing downtime of any serious computer to downtime of a PC isn't terribly meaningful. In this case, though, the interesting comparison is between mainframes and clusters.
But how much difference is there between mainframes and clusters? I remember one ATT "mainframe", for example, which was little more than a cluster of 3B2s with a bus that was faster than LAN technology. It was (if memory serves) 50 Mb/sec. It was faster than LAN technology then; not now.
As we step up to 10 Gb/s Ethernet, or as more expensive clusters use even faster/better/more expensive switching technologies (even IBM was doing this with RS6000s and some fast switch a few years ago), I'm curious what real differences will exist between cluster and mainframe.
At least as of now, I think it sufficient that we treat those looking to corrupt and destroy the government as criminals. Unfortunately, too many still treat these people as elected officials while others consider them merely incompetent.
The aftermath of Katrina should have shown the lie for how the current administration is protecting the citizens. What can we expect against an unexpected terrorist attack if something for which we'd days of warning can do so much damage with so little in the way of protection or even proper cleanup of the aftermath? Given what FEMA had become in the 90s and what's become of it now, it's a shock to me that anyone can associate handling of the "war on terror" in any successful way with the current administration.
And this isn't even discussing the embarassment of Tora Bora or spurious claims like:
Whereas members of al Qaida, an organization bearing responsibility for attacks on the United States, its citizens, and interests, including the attacks that occurred on September 11, 2001, are known to be in Iraq.
Still, I'm enough of an American, with a respect for due process, that I think hunted down and slaughtered a little strong. Ironic, in a way, that it is beliefs like that respect for due process that these destroyers wish to corrupt. I suppose that it is true that they're using our own beliefs and sense of honor against us.
I don't see where anyone here has mentioned that Gonzales has made the news recently a couple of times with ideas that seem designed to annoy people already opposed to the Bush Administration but which might pull along the weak-minded simpletons such as those blaming Iraq for the 2001/9/11 attack (see here for another example).
In other words, this looks designed to distract opponents and raise those approval ratings fast approaching the negative. I smell Rove in this.
Bush has protectionist leanings whenever it suits his purpose. The first was, I believe, the steel tarrifs. It wasn't the last, and I fully expect more as his party fights tooth and claw to stay in power lest we see some indictments in the very near future.
Those protectionist leanings show how *un*conservative this supposedly conservative administation truly is (if the exploding federal deficit isn't sufficient demonstration {8^).
Actually, this is about the situation I see in Firefox (plus Flashblocker) with a flash-heavy site. There's one I visit with some regularity as it has games and my son loves it (noggin.com perhaps?).
If I really cared, I'd "whitelist" the site in Flashblocker. But I don't care enough; it's too easy to click on the "play" button to see the flash.
So I don't see this being a big deal...except for those sites where the active content is something imposed upon the user who will therefore feel no desire to "play" it.
> You would think so. Starting with cookies, though, there's > always been a major component of web design and development > which hinges on deliberately obfuscating important events > from the user.
Still using cookies as an example, progress has been towards better "cookie privacy". Items like blocking 3rd party cookies by default, a clear "clear all information" button, limits which override cookie expiries, etc. all give the user more control over his/her privacy.
To add this "ping" feature w/o also providing control over its use to users is rather surprising since, otherwise, Firefox has been moving in the right direction.
This is not just surprising, but incredibly disappointing.
I thought I'd follow up with more Core3 information. The text-only install now does do an initial LVM setup. But it doesn't exploit RAID. So I want to change the setup...but there's still no way on the text-only side to add a new Volume Group (which one needs to do if one wants a group on a RAID set).
RHEL is not all that it can be as a server. My major peeve: incomplete support for headless installs. It is work to install over a serial port if you want features like LVM. Those exist only on the "graphical install" side of things.
There are also some "headless only" issues that should be addressed (ie. it should ask about the serial port's "magic key" initial state, given that there's no CTL, ALT or DEL key <grin>
In this, and in several other "headless" issues, I see no difference between Fedora and RHE.
That said, a lot is right. Installing headless gets the proper entries in grub.conf, for example. That was a pleasant surprise, and it works in both RHE and Fedora. Installing diskless using DHCP and TFTP is also quite easy (although I could wish for better documentation on this from Redhat; I've not yet figured out how to use kickstart in this environment).
Still, my key point: RHE still has room to evolve in server support, and I've seen little difference on these issues between RHE and Fedora.
As can I. I suspect that Google is taking an intelligent approach: Blocking IP blocks that are the sources of attacks. These would be networks infected with infected MSFT machines...and it is about time they suffer for their choices rather than forcing those costs upon the rest of us.
Reading the NYTimes article, noteworthy to me was that only the company in the middle (OptInBig) claimed to make money. The company which did the actual emailing and the company with the spamvertised web site were both (at least apparently) dead.
I've built numerous communication mechanisms over the years. Unless you're sufficiently lucky that your underlying environment and what you intend to build "match", a more abstracted underlying environment is going to give you trouble. The low level of the UNIX "stream" has the advantage that many different abstractions can be built over it.
This is actually quite similar to the nature and utility of a having a communication "stream" over IP.
You cite DCOP, JDBP, and such. These are Good Things, and powerful tools. But they've implicit assumptions. I'm glad to have them when I need them, but I'd rather not have them than not have the ability to drop down further when they don't fit.
UNIX has been around for a long time. What you see there is the result of decades of software evolution. But the origins aren't hidden. They're there for you to use...or ignore.
That they're there, though, is a part of the power of UNIX, and why it has so well withstood the test of time. It's a classic precisely because of its underlying general nature: it can fit a wide variety of models of thinking about software.
Perhaps because I've been working on the communication side of software for a while, I don't share your excitement over dotnet. Language independent data exchange formats have been around for a while. More interesting is a diversity of coordination models, but even this is not novel (although the patterns have been formally documented only relatively recently).
The real power of "communication" becomes apparent when one realizes that coordination is by far more important than merely moving data - or even information - around.
With respect to moving structured information over a stream: I built a marshalling mechanism for C data structures to a language independent data stream in about 1985 or 6. I was not the first, and I was not the last.
Slashdot Mirror
One of our mail servers was blocked by Comcast for a while. I'd a nice chat with one of their "technical supervisors" as a result.
I suggested that they offer a feedback solution like AOL's. He wanted to, but Comcast wasn't willing to spend the money on it. I then suggested that he let clients specify their own filtering rules. He wanted to do this too, but Comcast wasn't willing to spend the money on it.
We went around a few times this was before I realized that there was a dishonesty here. This isn't about protecting clients' mail boxes, despite what they're saying at Comcast. This is about cutting their costs. Spam requires handling. So they're just blocking anything that looks like spam to keep the mail server load lower than otherwise.
I suggested that some clients wanted to do their own filtering. That's why my servers will forward spam: we do have client specific filtering, and some clients want little or none on the server.
In other words, those Comcast clients that had expressedly told us to use little/no server-side filtering on their email were not being properly served by Comcast because Comcast is forcing its filtering upon those Comcast clients against their will. Worse, it could - in fact, does - cause these Comcast clients to miss messages they want to receive.
The answer to this was that anyone that was dependent upon email that way shouldn't be using a "consumer grade service". He implied that Comcast's business service didn't have this problem (although he wasn't sure).
So not only is Comcast looking to save money by cutting the load on their mail servers, they're downgrading their "consumer grade service" w/o cutting the price of this service. More, they are trying to push customers that complain into paying more.
I do wish I could be a monopoly too.
Your post is just MS bashing. MS gives customers what they want. Sometimes that's not good, because its not in the customers best interest ...
So the other posting was "MS bashing", but your comparison of MSFT to drug pushers, confidence artists, and the like isn't? Amusing.
You can explain how if it takes you twice as long to make sure it is easy to maintain in the future and will save money in the long run until you blue in the face. But what it comes down to is if your inital cost of development is less then the other guy.
This is why I love clients that have been previously burned by some code & cash developer [team]. They understand what the consequences of "low quality" mean, and are willing to pay to avoid it.
The more educated the client, the higher the demand for quality. It's just a shame that so many need to be burned to be educated.
Let me explain this as clearly as possible: business is about money.
You're young (or have been incredibly lucky in your experience).
[Or perhaps I've merely been unlucky.]
You speak of corporate strategy as if it is some mystical force beyond the ken of mere technologists. Yet you also accuse technologists of arrogance, which is nicely ironic.
Having some small experience along the technology/management divide, I can tell you that there are two issues you've missed. First, you missed the impact of the personal strategy of the manager(s) involved. I've seen projects impacted because of empire building, for example, or even just simple self-preservation. At a planning meeting for the installation of some software at a regional office, the manager of that office calmly explained that he'd support us completely until any capability of the new software threatened his job. At that point, he'd do everything he could to cause the project to fail.
Nicely aligned with corporate stragety there, eh?
I recall on another project having to waste several hours a week over a period of months defending a particular hardware choice. The choice we made was the obviously correct one, in that the resulting product was cheaper and more capable. But it did mean not exploiting an internal product.
We did consider the balance, but net it was still more profitable for the company to not use the internal product in this case. Yet meeting after meeting after meeting required that we explain this to the various "internal salespeople" for that product.
I eventually left that company over issues like this. The VP for whom I was working at the time tried to either convince me to stay or salve his own ego (I've never known which) by telling me a story. Back when he was an engineer, he'd had a similar choice: use an internal chip set or one from (I think) TI. The internal set cost on the order of a dollar per set; that one from outside cost on the order of pennies.
[Note: It's been 20 years; I may not recall those numbers accurately.]
Being the good company man, he chose internal.
I asked if the product had ever gone to market. It hadn't. It was too expensive to compete.
He told this to me, by the way, with a straight face. He actually [appeared to] still believe that he'd made the right choice.
Good corporate strategy there, eh?
I don't really mean to belittle that VP. He really was just following the strategy laid out for him. And, outside of issues like this, I recall him being a good manager. It was the strategy itself that was flawed in this case. Followed, it lost money and business for the company.
Of course, I'm assuming that the expression of this strategy was reaching us accurately. That need not have been the case. Perhaps some higher level manager (acting out of a personal strategy) warped it to serve his own interests. I'll never know.
Please don't misunderstand this to be a claim that technologists are never at fault for failures. I've other beefs with many of them. Working in the financial software world for a number of years, I was surprised at the number of software people that were completely disinterested in learning about the instruments whose behaviors they were modeling. If nothing else, it was *math*. What software engineer or programmer wouldn't enjoy the math of financial modeling?!?
I remain convinced that, at least in many cases, it was a form of arrogance that kept people disinterested.
Nevertheless, I felt I could not let your commentary slide by. Your idealistic view of a business as being honestly about profit is incomplete at best.
Doesn't code with lots of syntax errors turn out to have more bugs once it finally compiles? Why wouldn't the same be true for human language?
By all rights, engineers - people trained to work w/in system constraints or else - should be excellent writers. Rules of grammar and the like are conceptually no different than those of any engineering discipline. Prose is just another system.
Why this is so rarely the case I've never understood.
Today's scary thought: perhaps all governments that survive evolve into totalitarianism.
"Difference is the US doesn't deny this to its citizens."
Do you think that difference is worth a hill of beans?
I do. So does the current administration. That's why they're now looking to prosecute reporters that "hurt national security" by telling Americans the truth about our government.
Patriotism - in extreme forms often confused with blindness.
I think the word you're seeking is jingoism.
By contrast, the typical American supports human rights for both herself and others.
Unless the others include homosexuals that want to marry or pregnant women that don't want to be. Opposition to rights for these people can apparently be used to entice people to vote who'd otherwise not care enough about America to do so.
Anyone would be reviled in the US in public for forming an organization against civil rights for (for example) Blacks or the Irish. But homosexuals and pregnant women are apparently fair game. They're the real others today.
This is the power of "language survey" courses. As much as they provide an overview of the various languages, the real strength is that they teach the differences. How is LISP different from Java different from SML? What difference do those differences make, and how does it alter one's approach to problem solving.
Another advantage is that it provides access to a set of "problem solving" approaches that might not be "natural" in a given language but which might still be the best way to think about a problem.
I agree with those writing to skip the IDE in an introductory course. It's a distraction. Further, the advantages an IDE provides only truly appear with more complex software - software not likely to be written in an introductory course. Therefore, using the IDE that early (1) takes time away from the primary goal of the class and (2) forces the student to use a tool that provides little advantage.
Better to teach an IDE concurrent with more advanced programming projects that can really benefit from such a tool.
You cite the cost of that, and that is a valid concern. However, consider that no one proponent of this program will face the complexity. Using your example, each of the five networks will bill Google individually. There's some billing cost, but it should scale well: only paying "clients" are billed.
The complexity you cite will be faced instead by the content providers. In your example, it is Google that must track costs and issue payments to five different connectivity providers.
And Google already sees this as a Bad Idea. So while I agree that you're presenting a real cost, it's not a cost about which the proponents of this plan are going to be concerned.
However, my email client, my video player, and my web browser still run with the full privilege of my user account, when something less would be sufficient.
This is important, as many forms of malware (including that needed to build a 'bot) can be implemented w/o the requirement of root/superuser access. While the OS protecting itself is a Good Thing, this doesn't do anything to protect the computer itself against abuse (or to protect the Internet against abuse of this computer).
This is a fact too often missed during these discussions. And it's why we do need "least privilege", sandboxing, etc. for applications which execute untrusted content.
Further, comparing downtime of any serious computer to downtime of a PC isn't terribly meaningful. In this case, though, the interesting comparison is between mainframes and clusters.
But how much difference is there between mainframes and clusters? I remember one ATT "mainframe", for example, which was little more than a cluster of 3B2s with a bus that was faster than LAN technology. It was (if memory serves) 50 Mb/sec. It was faster than LAN technology then; not now.
As we step up to 10 Gb/s Ethernet, or as more expensive clusters use even faster/better/more expensive switching technologies (even IBM was doing this with RS6000s and some fast switch a few years ago), I'm curious what real differences will exist between cluster and mainframe.
Sorry, I know, it's your ejaculatory fantasy that everyone here is uber-geek elite
I think that this thread establishes that there aren't a lot of uber-geek elite to be found in tech support (at least in the US).
At least as of now, I think it sufficient that we treat those looking to corrupt and destroy the government as criminals. Unfortunately, too many still treat these people as elected officials while others consider them merely incompetent.
The aftermath of Katrina should have shown the lie for how the current administration is protecting the citizens. What can we expect against an unexpected terrorist attack if something for which we'd days of warning can do so much damage with so little in the way of protection or even proper cleanup of the aftermath? Given what FEMA had become in the 90s and what's become of it now, it's a shock to me that anyone can associate handling of the "war on terror" in any successful way with the current administration.
And this isn't even discussing the embarassment of Tora Bora or spurious claims like:
Whereas members of al Qaida, an organization bearing responsibility for attacks on the United States, its citizens, and interests, including the attacks that occurred on September 11, 2001, are known to be in Iraq.
Still, I'm enough of an American, with a respect for due process, that I think hunted down and slaughtered a little strong. Ironic, in a way, that it is beliefs like that respect for due process that these destroyers wish to corrupt. I suppose that it is true that they're using our own beliefs and sense of honor against us.
I personally find it funny that they would sacrifice their dignity for their perceived safety and future.
Not dignity but honor. And our Benjamin Franklin didn't think it funny, but he did warn us.
I don't see where anyone here has mentioned that Gonzales has made the news recently a couple of times with ideas that seem designed to annoy people already opposed to the Bush Administration but which might pull along the weak-minded simpletons such as those blaming Iraq for the 2001/9/11 attack (see here for another example).
In other words, this looks designed to distract opponents and raise those approval ratings fast approaching the negative. I smell Rove in this.
Bush has protectionist leanings whenever it suits his purpose. The first was, I believe, the steel tarrifs. It wasn't the last, and I fully expect more as his party fights tooth and claw to stay in power lest we see some indictments in the very near future.
Those protectionist leanings show how *un*conservative this supposedly conservative administation truly is (if the exploding federal deficit isn't sufficient demonstration {8^).
Actually, this is about the situation I see in Firefox (plus Flashblocker) with a flash-heavy site. There's one I visit with some regularity as it has games and my son loves it (noggin.com perhaps?).
If I really cared, I'd "whitelist" the site in Flashblocker. But I don't care enough; it's too easy to click on the "play" button to see the flash.
So I don't see this being a big deal...except for those sites where the active content is something imposed upon the user who will therefore feel no desire to "play" it.
> You would think so. Starting with cookies, though, there's
> always been a major component of web design and development
> which hinges on deliberately obfuscating important events
> from the user.
Still using cookies as an example, progress has been towards better "cookie privacy". Items like blocking 3rd party cookies by default, a clear "clear all information" button, limits which override cookie expiries, etc. all give the user more control over his/her privacy.
To add this "ping" feature w/o also providing control over its use to users is rather surprising since, otherwise, Firefox has been moving in the right direction.
This is not just surprising, but incredibly disappointing.
I thought I'd follow up with more Core3 information. The text-only install now does do an initial LVM setup. But it doesn't exploit RAID. So I want to change the setup...but there's still no way on the text-only side to add a new Volume Group (which one needs to do if one wants a group on a RAID set).
So close...so frustrating.
RHEL is not all that it can be as a server. My major peeve: incomplete support for headless installs. It is work to install over a serial port if you want features like LVM. Those exist only on the "graphical install" side of things.
There are also some "headless only" issues that should be addressed (ie. it should ask about the serial port's "magic key" initial state, given that there's no CTL, ALT or DEL key <grin>
In this, and in several other "headless" issues, I see no difference between Fedora and RHE.
That said, a lot is right. Installing headless gets the proper entries in grub.conf, for example. That was a pleasant surprise, and it works in both RHE and Fedora. Installing diskless using DHCP and TFTP is also quite easy (although I could wish for better documentation on this from Redhat; I've not yet figured out how to use kickstart in this environment).
Still, my key point: RHE still has room to evolve in server support, and I've seen little difference on these issues between RHE and Fedora.
As can I. I suspect that Google is taking an intelligent approach: Blocking IP blocks that are the sources of attacks. These would be networks infected with infected MSFT machines...and it is about time they suffer for their choices rather than forcing those costs upon the rest of us.
Reading the NYTimes article, noteworthy to me was that only the company in the middle (OptInBig) claimed to make money. The company which did the actual emailing and the company with the spamvertised web site were both (at least apparently) dead.
I'm not sure what this means, though.
I've built numerous communication mechanisms over the years. Unless you're sufficiently lucky that your underlying environment and what you intend to build "match", a more abstracted underlying environment is going to give you trouble. The low level of the UNIX "stream" has the advantage that many different abstractions can be built over it.
This is actually quite similar to the nature and utility of a having a communication "stream" over IP.
You cite DCOP, JDBP, and such. These are Good Things, and powerful tools. But they've implicit assumptions. I'm glad to have them when I need them, but I'd rather not have them than not have the ability to drop down further when they don't fit.
UNIX has been around for a long time. What you see there is the result of decades of software evolution. But the origins aren't hidden. They're there for you to use...or ignore.
That they're there, though, is a part of the power of UNIX, and why it has so well withstood the test of time. It's a classic precisely because of its underlying general nature: it can fit a wide variety of models of thinking about software.
Perhaps because I've been working on the communication side of software for a while, I don't share your excitement over dotnet. Language independent data exchange formats have been around for a while. More interesting is a diversity of coordination models, but even this is not novel (although the patterns have been formally documented only relatively recently).
The real power of "communication" becomes apparent when one realizes that coordination is by far more important than merely moving data - or even information - around.
With respect to moving structured information over a stream: I built a marshalling mechanism for C data structures to a language independent data stream in about 1985 or 6. I was not the first, and I was not the last.