"When people watch fictional shows, most of them want to see shows about human drama. Engineering isn't about human drama"
Self-fullfilling profecy.
Engineering is *a lot* about human drama. Basically all Humankind is is around engineering.
Tell me you can't produce a dramatic film about basically *any* big engineering challange of Humanity. The pyramids, the cathedrals, the Panama Channel, the Golden Gate, the first skyscrappers... you name it.
But since "people don't want that", the entertaiment industry don't give that and since they don't give that, they are not expected to give that.
Posting on forums is a social trend. I never doubted that web sites won the fight about this years ago, but that was not my point, my point is that NNTP is technically vastly superior to its web-based lessen brothers. Slashdot is a perfect case for this: * Try to follow a thread with more than five/six indentantion levels. * Try to track cites within messages. * Try to follow on a story once it left the front page. * Try to do advanced message scoring -heck try to do anything that was not in the mind of the site owners. * Try to retain messages/threads that specially interest you.
Mind you, some things aforementioned are somehow doable on Slashdot, only at great pain, but are basically trivial with a NNTP client.
"The real problem is that is the perception."
Of course it is. That's why I said that NNTP is better *on technical grounds*.
How said he is taking responsibility (in the sense of "yes, it's my fault")?
There are two kinds of responsibility-related resignations: 1) As a way to say "I failed, I don't deserve this position". 2) As a way to say "I tried to do my job but the higher ups don't allow me to do it properly: I won't continue under these circumnstances".
No where in the article nor the links there's indication about what's the case here.
"On the other hand, there's a barely anyone using Usenet binary groups for legitimate purposes"
Yes, and I certainly would support closing the alt.binary hierarchy (or at least not pressing for its maintenance): not only NNTP never was a good tool to support binary files but there are much better tools really oriented from the ground up to that today.
It's only that they are closing service to *all* groups, not only alt.binary.
Not only a single UI but the specific UI that suits my needs/preferences, which doesn't need to be the same that fits others.
"and a single searchable archive of all your posts."
And local retention if I want to and exactly the way I want to, and the ability to score messages by the criteria better fits me, and a logical and common hierarchy to find/subscribe to info...
I honestly believe that the only way to say that web-based service is a better tool for on-line discussion is not knowing about the alternatives.
"For most people, web-based services are a better medium for online discussions."
It is not. It only seems that way at first glance. On the part of the user/consumer is has a slightly higher entry effort than a forum because in the latter you see on the spot what's happening at the price of much lower usability*1 and control *forever* -a pity, but quite a common trend. On the side of the producer/site owner, it allows easier control and add revenue by positioning non-solicited info along the messages. No one of them are technical advantages and even less so advantages for the end user.
"NNTP was never a particularly good protocol for distributing large binaries"
True, nor it was meant for that. Binaries were always an after-thought and side-stepping on the main feature: hierarchically organized decentralized-controlled discussion forums.
*1 Horrid nesting, difficult to follow long threads, inability to score, tag, selectively retain, etc. client side...
"Technology has moved on, and Usenet is an anachronism."
So, please, can you explain to me what's the better technology that arose that made NNTP an anachronism? Because I honestly say I don't know the current technology that is better than NNTP doing its stuff on technical grounds.
"Are you suggesting 99% of all liquor is consumed by drunk drivers ?"
No. He is suggesting 99% of drunk drivers bought it to liquor companies and that 100% of murderers using guns bought it from a gun-producing company.
Which certainly is a good analogy with regards of NNTP and copyright-infringent content (which, by the way, is *not* copyright-infringent content at least on some EU countris, ACTA notwithstanding).
"And you think somehow, that I will give a shit about my data."
Yes. It was at 4AM and it was just a big damn fire, so nobody is injured. The first week is a nightmare, yes, but then, you recall your insurance and hire a new office and then, what? Where's your customers data, your financial records... your everything?
Small business tend to undervaluate how dependant they are on their data (except for the from-time-to-time cry for help from somebody "please, how can I recover my hard disk? If I can't do it, I'll have to close my business -no, I don't have any backup, of course").
Security is not easy. Not in the physical world, not in the intertubes. And people don't really worry about security (not in advance, at least), so they deem to be "too hard".
"No, Occam's Razor says absolutely nothing about truth-value"
Nor did I attribute it. Truth is a matter of Maths, not Science.
I thought that using the verb "bet" made clear I was not going but informally into epystemology but whatever.
"Which is "true"--Euclidean Geometry or Riemannian Geometry? [...] Occam's Razor indicate is better to "use" for the general case [...] Euclidean"
Again, that's Maths, not Science, but the razor can be equally used here and do not say what you say it says.
The first part of the dictum is "entia non sunt multiplicanda", which may point to Euclides (in fact, it doesn't, it might point to absolute geometry, which is not the same, but I understand what you mean), but the second part adds "praeter necessitatem", "...beyond needs", so if you need to work over a sphere's surface, your needs makes Riemman's the proper choice. What it does too, is favour the minimal number of axiomes to be used. As a general matter, it begs the question "is this object needed to explain what we came to explain"? So, in the case of cosmology, "is the object called "god" needed to explain this other object called universe? because if we can explain Universe without adding the object called god, we'd better leave it alone".
"The misapplication of this principle is amazingly widespread..."
Or not.
"entia non sunt multiplicanda praeter necessitatem"
So you may choose between "a universe" and "a universe *and* a god to create it". As long as there's no need of god for the universe to be, Occam bets for the former option.
"A truly scientific worldview would lead to opinions such as: we don't know, there is no proof one way or the other."...therefore, till new proofs appear, we'll stick to the simplest explanation, the one without the imaginary friend, that is.
"The problem is that if Red Hat dies, there will be no new version of RHEL or CentOS."
The problem is that if "Company X" dies, there will be no new products and services from "Company X".
How is that any news? Red Hat decided a business strategy, with its pros and cons, and it is their business strategy. Am I be the one that will support their business strategy? I don't think so, since I'm not a Red Hat CxO.
They try to do their best and I try to do my best. That's called capitalism. Do you want a given company class not to fail? It's easy: make it a State-owned one. Do you want companies that don't deserve people's money from the point of view of those people as they vote with their wallets to still have money guaranteed? Press for socialism instead of capitalism.
"Buying Red Hat is the better long-term solution"
Maybe you are right but, as I think Keynes stated, in the long-term we are all dead.
"if you don't assume you're the only entity makin decisions"
But the fact is that I *am* the only entity making decisions *I* can control.
"And that's where stellar support can make a huge difference."
Regarding software configuration I can do as good as Red Hat, thank you.
Regarding user-level bugs, Red Hat can do liminary better than me and my "usual" extended support team (both in-house and externalized), and only from time to time (i.e. GFS).
Regarding kernel-level bugs, well, a server is going crazy... it might be the SCSI card flaking or maybe it is the SCSI kernel driver, or its interaction with the kernel or with other components. Do you think Red Hat by itself is going to cover the situation timely enough? More importantly: is it going to do timely enough that I don't need other security layers in order to guarantee my internal SLA (i.e, high avaliability, at least two hardware vendors for critical stuff, etc.)? And if I still need those other layers, does paying for Red Hat support too pay off?
Again:
"you have to weigh the costs of downtime versus the cost of support"
Not. You have to weigh the *differential* downtime cost (as if it was an easy task except for the trivial case of wholly managed services ala IBM) versus the support cost.
For the most part, OS-level support, specially when talking about open source OSs, is not because of the financial case but because of the CYA one.
"The boss doesn't believe in support. CentOS is a product with no support. Do it, and if shit hits the fan you have your big "I told you so""
And you surely don't forget to add to your "I told you so" the exact way in that having payed for Red Hat support would have avoided or at least alleviated the effect of the shit hitting the fan.
Because without that, you still have no damn case.
"there are products out there, or drivers for said products, which will ONLY work on a RHEL box because of RPM package dependencies or library linking to libraries of different names/etc."
Will it's probably the case that some vendors will refuse to support their products on CentOS instead of Red Hat, that's a political decision, not technically-based.
AFAIK CentOS is supposed to be binary-compatible and certainly will have the same package dependencies/libraries names than the "genuine" Red Hat.
Then you go mixing apples with oranges about RHEL, CentOS, Oracle's EL, Ubuntu... So, please, can you offer an explicit example or are you just spreading FUD?
"Hey, if we strip all the copyrighted stuff out we can just take what we want and not have to pay RH shit! We'll save a bundle!"
Well, and the "no so nice" part is?
Red Hat decided on their own way to do business. Such a way included not developing an OS from start but instead using an OS with a license that allowed them to package it and throw a brand, a marketing campaign and a support business but it has a cost Red Hat was willing to accept: that others could do the same.
The end result is that Red Hat pushes money at it because it works for them, CentOS rebrands the software because it works for them, and I as a user have a choice that fits me. The day each respective choice works for the given agent no more is the day they'll change boats to look for greener coasts.
But that's the basis of free market, now, isn't it?
Slashdot Mirror
"Let's see you make such a film, that's actually interesting and appealing to non-engineers"
I'm not in that career but I can give an example: Apollo XIII.
Human drama around engineers doing their stuff.
"When people watch fictional shows, most of them want to see shows about human drama. Engineering isn't about human drama"
Self-fullfilling profecy.
Engineering is *a lot* about human drama. Basically all Humankind is is around engineering.
Tell me you can't produce a dramatic film about basically *any* big engineering challange of Humanity. The pyramids, the cathedrals, the Panama Channel, the Golden Gate, the first skyscrappers... you name it.
But since "people don't want that", the entertaiment industry don't give that and since they don't give that, they are not expected to give that.
"I'm fairly sure Windows is a general purpose OS with no one philosophy"
Then, IMNSHO your are wrong.
"It has a market share of over 90% - every kind of person uses Windows."
Flawed argument: almost every kind of person uses Windows... in basically the same way.
"So says the guy posting on slashdot..."
Posting on forums is a social trend. I never doubted that web sites won the fight about this years ago, but that was not my point, my point is that NNTP is technically vastly superior to its web-based lessen brothers. Slashdot is a perfect case for this:
* Try to follow a thread with more than five/six indentantion levels.
* Try to track cites within messages.
* Try to follow on a story once it left the front page.
* Try to do advanced message scoring -heck try to do anything that was not in the mind of the site owners.
* Try to retain messages/threads that specially interest you.
Mind you, some things aforementioned are somehow doable on Slashdot, only at great pain, but are basically trivial with a NNTP client.
"The real problem is that is the perception."
Of course it is. That's why I said that NNTP is better *on technical grounds*.
"Kudos to him for taking responsibility"
How said he is taking responsibility (in the sense of "yes, it's my fault")?
There are two kinds of responsibility-related resignations:
1) As a way to say "I failed, I don't deserve this position".
2) As a way to say "I tried to do my job but the higher ups don't allow me to do it properly: I won't continue under these circumnstances".
No where in the article nor the links there's indication about what's the case here.
"On the other hand, there's a barely anyone using Usenet binary groups for legitimate purposes"
Yes, and I certainly would support closing the alt.binary hierarchy (or at least not pressing for its maintenance): not only NNTP never was a good tool to support binary files but there are much better tools really oriented from the ground up to that today.
It's only that they are closing service to *all* groups, not only alt.binary.
"USNET's other advantages are a single UI "
Not only a single UI but the specific UI that suits my needs/preferences, which doesn't need to be the same that fits others.
"and a single searchable archive of all your posts."
And local retention if I want to and exactly the way I want to, and the ability to score messages by the criteria better fits me, and a logical and common hierarchy to find/subscribe to info...
I honestly believe that the only way to say that web-based service is a better tool for on-line discussion is not knowing about the alternatives.
"For most people, web-based services are a better medium for online discussions."
It is not. It only seems that way at first glance. On the part of the user/consumer is has a slightly higher entry effort than a forum because in the latter you see on the spot what's happening at the price of much lower usability*1 and control *forever* -a pity, but quite a common trend. On the side of the producer/site owner, it allows easier control and add revenue by positioning non-solicited info along the messages. No one of them are technical advantages and even less so advantages for the end user.
"NNTP was never a particularly good protocol for distributing large binaries"
True, nor it was meant for that. Binaries were always an after-thought and side-stepping on the main feature: hierarchically organized decentralized-controlled discussion forums.
*1 Horrid nesting, difficult to follow long threads, inability to score, tag, selectively retain, etc. client side...
"Technology has moved on, and Usenet is an anachronism."
So, please, can you explain to me what's the better technology that arose that made NNTP an anachronism? Because I honestly say I don't know the current technology that is better than NNTP doing its stuff on technical grounds.
"Are you suggesting 99% of all liquor is consumed by drunk drivers ?"
No. He is suggesting 99% of drunk drivers bought it to liquor companies and that 100% of murderers using guns bought it from a gun-producing company.
Which certainly is a good analogy with regards of NNTP and copyright-infringent content (which, by the way, is *not* copyright-infringent content at least on some EU countris, ACTA notwithstanding).
"Seriously?"
Yes.
"And you think somehow, that I will give a shit about my data."
Yes. It was at 4AM and it was just a big damn fire, so nobody is injured. The first week is a nightmare, yes, but then, you recall your insurance and hire a new office and then, what? Where's your customers data, your financial records... your everything?
Small business tend to undervaluate how dependant they are on their data (except for the from-time-to-time cry for help from somebody "please, how can I recover my hard disk? If I can't do it, I'll have to close my business -no, I don't have any backup, of course").
"when it comes to trust, you need a centralized authorit"
Probably you are right.
But then, when it comes to trust, you can never trust a centralized authority.
"OpenPGP is too hard for normal people"
And that's exactly the point.
Security is not easy. Not in the physical world, not in the intertubes. And people don't really worry about security (not in advance, at least), so they deem to be "too hard".
"Who says you have to live in a Third World country for there to be data caps?"
He implied that having data caps is a sign of thirdworldliness no matter the country, which I concur.
"No, Occam's Razor says absolutely nothing about truth-value"
Nor did I attribute it. Truth is a matter of Maths, not Science.
I thought that using the verb "bet" made clear I was not going but informally into epystemology but whatever.
"Which is "true"--Euclidean Geometry or Riemannian Geometry? [...] Occam's Razor indicate is better to "use" for the general case [...] Euclidean"
Again, that's Maths, not Science, but the razor can be equally used here and do not say what you say it says.
The first part of the dictum is "entia non sunt multiplicanda", which may point to Euclides (in fact, it doesn't, it might point to absolute geometry, which is not the same, but I understand what you mean), but the second part adds "praeter necessitatem", "...beyond needs", so if you need to work over a sphere's surface, your needs makes Riemman's the proper choice. What it does too, is favour the minimal number of axiomes to be used. As a general matter, it begs the question "is this object needed to explain what we came to explain"? So, in the case of cosmology, "is the object called "god" needed to explain this other object called universe? because if we can explain Universe without adding the object called god, we'd better leave it alone".
"The misapplication of this principle is amazingly widespread..."
Or not.
"entia non sunt multiplicanda praeter necessitatem"
So you may choose between "a universe" and "a universe *and* a god to create it". As long as there's no need of god for the universe to be, Occam bets for the former option.
"A truly scientific worldview would lead to opinions such as: we don't know, there is no proof one way or the other." ...therefore, till new proofs appear, we'll stick to the simplest explanation, the one without the imaginary friend, that is.
Occam's razor, they call it.
"The problem is that if Red Hat dies, there will be no new version of RHEL or CentOS."
The problem is that if "Company X" dies, there will be no new products and services from "Company X".
How is that any news? Red Hat decided a business strategy, with its pros and cons, and it is their business strategy. Am I be the one that will support their business strategy? I don't think so, since I'm not a Red Hat CxO.
They try to do their best and I try to do my best. That's called capitalism. Do you want a given company class not to fail? It's easy: make it a State-owned one. Do you want companies that don't deserve people's money from the point of view of those people as they vote with their wallets to still have money guaranteed? Press for socialism instead of capitalism.
"Buying Red Hat is the better long-term solution"
Maybe you are right but, as I think Keynes stated, in the long-term we are all dead.
"if you don't assume you're the only entity makin decisions"
But the fact is that I *am* the only entity making decisions *I* can control.
" As you might probably guess, my deployments never result in any four-letter-words being uttered."
Oh, that's good. That means you couldn't be hitted by the DNS TTL (bug? feature?) from Amazon, or could you?
"Sure it can."
But of course it can. The point is: it *is*?
"And that's where stellar support can make a huge difference."
Regarding software configuration I can do as good as Red Hat, thank you.
Regarding user-level bugs, Red Hat can do liminary better than me and my "usual" extended support team (both in-house and externalized), and only from time to time (i.e. GFS).
Regarding kernel-level bugs, well, a server is going crazy... it might be the SCSI card flaking or maybe it is the SCSI kernel driver, or its interaction with the kernel or with other components. Do you think Red Hat by itself is going to cover the situation timely enough? More importantly: is it going to do timely enough that I don't need other security layers in order to guarantee my internal SLA (i.e, high avaliability, at least two hardware vendors for critical stuff, etc.)? And if I still need those other layers, does paying for Red Hat support too pay off?
Again:
"you have to weigh the costs of downtime versus the cost of support"
Not. You have to weigh the *differential* downtime cost (as if it was an easy task except for the trivial case of wholly managed services ala IBM) versus the support cost.
For the most part, OS-level support, specially when talking about open source OSs, is not because of the financial case but because of the CYA one.
"The boss doesn't believe in support. CentOS is a product with no support. Do it, and if shit hits the fan you have your big "I told you so""
And you surely don't forget to add to your "I told you so" the exact way in that having payed for Red Hat support would have avoided or at least alleviated the effect of the shit hitting the fan.
Because without that, you still have no damn case.
"If you're deploying anything straight to production without testing that exact thing somewhere else first, you deserve whatever you get."
So you do own an exact replica of your production environment, including all expensive hardware, load and size for staging purposes?
"there are products out there, or drivers for said products, which will ONLY work on a RHEL box because of RPM package dependencies or library linking to libraries of different names/etc."
Will it's probably the case that some vendors will refuse to support their products on CentOS instead of Red Hat, that's a political decision, not technically-based.
AFAIK CentOS is supposed to be binary-compatible and certainly will have the same package dependencies/libraries names than the "genuine" Red Hat.
Then you go mixing apples with oranges about RHEL, CentOS, Oracle's EL, Ubuntu... So, please, can you offer an explicit example or are you just spreading FUD?
"The question is not how much support costs. The question is how much is DOWNTIME going to cost the company?"
You say it as if paying Red Hat's support would magically lower DOWNTIME when compared to using CentOS.
Now, is that the case?
"Hey, if we strip all the copyrighted stuff out we can just take what we want and not have to pay RH shit! We'll save a bundle!"
Well, and the "no so nice" part is?
Red Hat decided on their own way to do business. Such a way included not developing an OS from start but instead using an OS with a license that allowed them to package it and throw a brand, a marketing campaign and a support business but it has a cost Red Hat was willing to accept: that others could do the same.
The end result is that Red Hat pushes money at it because it works for them, CentOS rebrands the software because it works for them, and I as a user have a choice that fits me. The day each respective choice works for the given agent no more is the day they'll change boats to look for greener coasts.
But that's the basis of free market, now, isn't it?