It really depends on what you call 'harder'. If it's the amount of time and sheer manpower spent, then yes, support is much harder than writing the software. However, if you equate 'harder' with pure ability and availability of people being capable of doing it, then writing a good piece of software is much much harder than supporting something. Any company can support and document any piece of shit by simply hiring enough practically random people and train them to do rote support. Most 'support' follows this model.
It's expensive, it's useless, but it sure is professional.
Very few companies are however capable of hiring and keeping the talent that is necessary to write a high-quality piece of software. Writing good software is hard and only a few people are capable of doing it. Writing an unmaintanable mess that barely meets the specifications is easy, that's what code-monkeys do.
Just looking at the quality of current 'enterprise class' software (SAP, Oracle, Siebel) supports the case that writing even decent software, let alone high quality, is very hard indeed. In general, professional organizations do not seem to be capable of it as they consider their core talent as an interchangeable resource. Hence lack of quality and even more dependence on support.
I've worked for large scale service organizations (specialized in SAP implementations), small scale software vendors (creating their own shit using a set of more-or-less talented developers) and also dabbed into open source. The big difference between open source and the commercial integrators/developers is that OSS is *not* done when the system appears to work. As the people like their code, they will keep on tinkering with it until the core of the system is nearing perfection. Documentation and supports suffers from this. However, commercial software has much less emphasis on getting the core right, once it works and doesn't have too many bugs in it, it will survive many releases unchanged. There is no drive to perfection as there's no buck to be made with that. The drive shifts to sales and support structures and incremental changes, mainly in the client-visible parts of the application. No money is ever spent anymore to make the core code better, unless it is explicitly necessary because the poor quality starts to hurt the bottom line.
The net result is that over the years, OSS evolves to a realatively undocumented piece of very robust and scalable code, while non-OSS software becomes an ungodly bloated mess with twenty books of documentation and a (customer paid) full-scale support team, and it still goes down from time to time. What sort of software would you trust your business to?
I'm 36, and I consider 18 year olds to be no more than children. Almost there, but still children. From 25 year onward I start to take people 100% seriously. Don't really care that much about the opinion of kids younger than that.
Okay, to write maintanable code, it is a great help if you have the mathematical concept of preconditions, postconditions and invariants handy. Especially the latter one is overlooked. If you work in an OO shop and you have a class that is misbehaving, check if it has a well-defined invariant that specifies unambiguously the interrelationships of the data members. Write down this invariant. If you can't do it, you need to refactor your code into a couple of classes that do have an invariant.
Then check, that (a) the constructor establishes the invariant, (b) every method leaves the invariant intact after exit, and (c) that the method it is actually necessarily a method, i.e., it needs to lift the invariant temporarily to get its work done. If (c), make the method a regular function with no more than public access to the class. Do this religuously and you will spot the error promptly and as a side-effect end up with a robust program (and probably only 20% of your code in objects).
Preconditions, postconditions and invariants are mathematical concepts that describe your code. Use them and you're writing better code. It would even be better to actually prove them, but that would mean your code would not get out of the door in time. When you get into trouble though, making this analysis will give you a clear view where the problems are located in your program. Math (in this case formal reasoning about your code) in action: guidelines for what a class is supposed to do, when things are members and when not, testable descriptions of your code, and a way to pinpoint errors.
The reason why mathematics should be taught in highschool is that people should gain some concept of logic, which is useful no matter where you're headed, and by proving propositions and theorems, one eventually gains an incredible grasp of logic. This isn't currently done though.
But why o why doesn't logic get taught in school? The concepts of propositional logic, and even predicate calculus are sufficiently simple to even learn in the very first years of high school. You can apply this to language and even teach those 'alpha' kids to recognize invalid reasoning. No, what we get at first is 'math', primarily aimed at calculations on the real axes, not the root of math, valid and exact reasoning. Math is necessary, but without logic it lives in a void.
By not teaching logic as a proper subject in school, the politicians have already won!
Replace 'maths' with 'calculus' in your post and you have a point. In reality, as a programmer you're applying something called 'discrete math'. Discrete math studies diverse things as groups, sets, modulo arithmetic, trees, graphs, hashing, lists, arrays, floating point arithmetic, complexity analysis, information theory and more, much more. You might have encountered some of this discrete math in your 15 year programming career.
If I have seen further than others, it is because I have stood on the shoulders of giants--DUMB GIANTS THAT SUCK!
Pretty bad. If you want to insult a man, do it like Sir Isaac Newton did. His remark about the giants was not originally his, but what was original was that it was send to Robert Hooke. Hooke had claimed repeatedly that Newton stole much of his work. The real nastiness of Newton's letter to Hooke can be found in the fact that Hooke is often described as a dwarf.
And thus, instead of working towards it, you randomly start locking up people, dispense your justice system and invade other countries, antagonizing both friend and foe? Now that's logical!
It surely seems that the only war the US is waging at this time with any remote chance of success is its 'war on rationality'. The US citizens are getting pretty good at fighting it.
As for the US, they're now (by far?) the biggest. But even if they were twice as big, US anti-trust laws are no longer enforced, anything goes. The EU has sharper teeth, but given that SAP is still globally larger than Oracle, and in Europe dominating, I doubt they'll get a stink from that end of the world. So all in all, no roadblocks as far as I can see.
Imagine that they offer you a stock model in which you feed the economic situation of before 1999, and let it run using that information alone. So no peeks at any data after 99. Further suppose that you can run this model to predict a particular stock or index until 2005 and it is makes good predictions: it predicts the bubble, it predicts the burst and all that comes after it. It also predicts the national debt of the US and of each country important enough to influence the global economy. All that data can be read out of the internal state of the model.
This model would then be trustworthy as it is a complete model of the economic reality + stockholder psychology that predicts well for long periods of time.
This is the kind of model the GP is talking about. It's internal state is some information about the climate, and the test is to see if it can update its state consistenly only given initial condition. You can test it for various things. It generates data, it doesn't merely predict. No stock model exists that can do this for even a few days accurately. To be able to do this, the model should have a fairly complete internal state that mimics economy: distribution of wealth for individuals, institutions, nations. A good grasp of the psychology that these actors have, and it should be the case that the economy is non-chaotically deterministic. This last point is probably untrue. You don't see real models for the stocks market but merely pattern recognizers, smart interpolators. There is no theory that says that patterns necessarily repeat, thus you see the overfitting behaviour of such 'models'.
That's absolutely right, but I haven't seen anyone yet questioning the Corps of Engineers about their decision that a 1 in 200 years event was more than enough to protect against.
I'm Dutch, I live below sea-level. We don't get hurricanes, but we do know the danger we're in. We aim to be overwhelmed by events that happen 1 in 10,000 years (our events can be big, study our geography, note the location of the Channel and also note that not all water can go through it). Some investments we made have been ridiculed by the US corps of engineers. We spent appr. 40 billion dollars in 45 years (!) to prevent an event of a scale that is larger than the one that happened in NO. Imagine: Rotterdam, largest harbour in the world, 10 million people living below or closely at sea-level. I hope you get the picture.
What was the Corps of Engineers thinking in the US? That a 1 in 200 years event wouldn't happen in their lifetime? Well, given an expected age of 80 years, 40% chance that is does. For a single event! This is a situation that was known and budgetted for for at least 40 years. And now the Americans are all in tears!
People, you have willfully neglected the risks for an event like this, probably because it's a cheap way to run a government. Now you are hurting. Please start to consider costs and benefits, examine the cost that would have incurred if the levees were brought up to par in the sixties, compare it with the costs now (excluding the unnecessary loss of life), and make sure it won't happen again.
Ah, who am I kidding. The US as a noob nation is incable of looking ahead more than a few years at a time. This in another fact that has been proven time and time again. When will you learn to consider life more important than the almighty buck.
Joe Blow should know better than to run a software business next to his 65 hour work week. Oh, Open Source you say? You mean like the kind that has no warranty, no fitness for a particular purpose and no monetary value whatsoever being exchanged?. Don't worry, Joe Blow hasn't sold a thing to anyone, he is safe.
Ok, didn't realize Microsoft is 100% floating and apparently never read it before (so keep on repeating). Nevertheless, for Microsoft to fail quickly, the PC market should almost have to evaporate overnight. This is simply not going to happen. Maybe that Microsoft is marginalized in one/two decades, but that's about the best I can imagine.
What do you think the shareholders use to track their investments?
Not really true. IBM might have been a big player in PC space, yet PC space was insignificant at the time. It's like saying Microsoft is in the market of fabricating mice. True, but hopelessly inadequate.
A more apt comparison would be to say that IBM was a huge player in the mainframe market. And where are they now? They're still a huge player in the mainframe market. It is just that this market has slowly eroded, and IBM is slowly changing to accommodate this. Likewise, only when the PC market will dwindle does Microsoft have anything to fear. Finally, 40 billion dollars go a long way for living through hard times. Don't count on Microsoft going away in the next few decades.
He's no asshole, they pissed him off and he decided to play a prank on the suckers at ruckers. He did, and it was funny. Hence the presence of the story on Slashdot.
Unfortunately it seems that slashdot these days is mostly populated by morally indignant, self-righteous jerks that lack any semblance of a sense of humour. The situation doesn't call for a heavy-handed set of arguments about the relative moralities of the parties involved. The situation is simply funny, laugh.
Still I think taking the toy away is the better course of action. Access restrictions on the routers might also be something to contemplate in such an environment, but maybe the schools really aren't up to such a technological challenge. But going through heavy duty legal stuff to make up for their own incompetence to handle their pupils doesn't really sit well with me.
Actually, we do try to keep the passwords in sync more or less. It usually takes around three guesses to find it. The point is still that we're a small outfit and we don't have financial accounts to protect. The first step in security is determining what needs protection. In our outfit, internal access is open, though external access is purely through a stored public key on the access point. The private keys are personal.
Not sure why this scored funny, but I think it's an interesting thought for open source software. What would happen when open-source software would just start using generic names that absolutely cannot be trademarked. Given the nature of open source, having or not having a trademark is fairly irrelevant, so why not use this. I can see it now:
Mambo renames to webportal
firefox renamed to browser
openoffice renamed to:
word processor
spread sheet
etc.
linux (and(!)/or BSD) renamed to 'operating system'
And of course in a development environment, where you expect that you can sit behind someone else's machine in any place (and without a good LDAP implementation in place due to IT incompetence), we actually *require* everyone to put their password under their keyboard (or actually on a post-it note on the monitor when they go on holidays).
Note that we're working on windows machines, where it takes a good half day to create a workable development environment if you log in as a new user. Re-using a co-workers environment thus saves hours. As for outside intrusions, we have a lock on the door and an alarm as well.
Slashdot Mirror
Very few companies are however capable of hiring and keeping the talent that is necessary to write a high-quality piece of software. Writing good software is hard and only a few people are capable of doing it. Writing an unmaintanable mess that barely meets the specifications is easy, that's what code-monkeys do. Just looking at the quality of current 'enterprise class' software (SAP, Oracle, Siebel) supports the case that writing even decent software, let alone high quality, is very hard indeed. In general, professional organizations do not seem to be capable of it as they consider their core talent as an interchangeable resource. Hence lack of quality and even more dependence on support.
I've worked for large scale service organizations (specialized in SAP implementations), small scale software vendors (creating their own shit using a set of more-or-less talented developers) and also dabbed into open source. The big difference between open source and the commercial integrators/developers is that OSS is *not* done when the system appears to work. As the people like their code, they will keep on tinkering with it until the core of the system is nearing perfection. Documentation and supports suffers from this. However, commercial software has much less emphasis on getting the core right, once it works and doesn't have too many bugs in it, it will survive many releases unchanged. There is no drive to perfection as there's no buck to be made with that. The drive shifts to sales and support structures and incremental changes, mainly in the client-visible parts of the application. No money is ever spent anymore to make the core code better, unless it is explicitly necessary because the poor quality starts to hurt the bottom line.
The net result is that over the years, OSS evolves to a realatively undocumented piece of very robust and scalable code, while non-OSS software becomes an ungodly bloated mess with twenty books of documentation and a (customer paid) full-scale support team, and it still goes down from time to time. What sort of software would you trust your business to?
I'm 36, and I consider 18 year olds to be no more than children. Almost there, but still children. From 25 year onward I start to take people 100% seriously. Don't really care that much about the opinion of kids younger than that.
Preconditions, postconditions and invariants are mathematical concepts that describe your code. Use them and you're writing better code. It would even be better to actually prove them, but that would mean your code would not get out of the door in time. When you get into trouble though, making this analysis will give you a clear view where the problems are located in your program. Math (in this case formal reasoning about your code) in action: guidelines for what a class is supposed to do, when things are members and when not, testable descriptions of your code, and a way to pinpoint errors.
gah, should have previewed the subject, corrected above.
But why o why doesn't logic get taught in school? The concepts of propositional logic, and even predicate calculus are sufficiently simple to even learn in the very first years of high school. You can apply this to language and even teach those 'alpha' kids to recognize invalid reasoning. No, what we get at first is 'math', primarily aimed at calculations on the real axes, not the root of math, valid and exact reasoning. Math is necessary, but without logic it lives in a void.
By not teaching logic as a proper subject in school, the politicians have already won!
Sine(alpha) = Opposite/Hypothenuse
And we're interested in the angle alpha. How do we get it out? Simple:
alpha = ArcSine( Opposite/Hypothenuse)
And voila, there's your trancendental function.
All in all, math != calculus.
Pretty bad. If you want to insult a man, do it like Sir Isaac Newton did. His remark about the giants was not originally his, but what was original was that it was send to Robert Hooke. Hooke had claimed repeatedly that Newton stole much of his work. The real nastiness of Newton's letter to Hooke can be found in the fact that Hooke is often described as a dwarf.
And thus, instead of working towards it, you randomly start locking up people, dispense your justice system and invade other countries, antagonizing both friend and foe? Now that's logical!
It surely seems that the only war the US is waging at this time with any remote chance of success is its 'war on rationality'. The US citizens are getting pretty good at fighting it.
As for the US, they're now (by far?) the biggest. But even if they were twice as big, US anti-trust laws are no longer enforced, anything goes. The EU has sharper teeth, but given that SAP is still globally larger than Oracle, and in Europe dominating, I doubt they'll get a stink from that end of the world. So all in all, no roadblocks as far as I can see.
This is the kind of model the GP is talking about. It's internal state is some information about the climate, and the test is to see if it can update its state consistenly only given initial condition. You can test it for various things. It generates data, it doesn't merely predict. No stock model exists that can do this for even a few days accurately. To be able to do this, the model should have a fairly complete internal state that mimics economy: distribution of wealth for individuals, institutions, nations. A good grasp of the psychology that these actors have, and it should be the case that the economy is non-chaotically deterministic. This last point is probably untrue. You don't see real models for the stocks market but merely pattern recognizers, smart interpolators. There is no theory that says that patterns necessarily repeat, thus you see the overfitting behaviour of such 'models'.
Well, given the originality of Wolfram's work, I'm pretty confident that "A New Kind of Sex" will simply promote abstinence.
I'm Dutch, I live below sea-level. We don't get hurricanes, but we do know the danger we're in. We aim to be overwhelmed by events that happen 1 in 10,000 years (our events can be big, study our geography, note the location of the Channel and also note that not all water can go through it). Some investments we made have been ridiculed by the US corps of engineers. We spent appr. 40 billion dollars in 45 years (!) to prevent an event of a scale that is larger than the one that happened in NO. Imagine: Rotterdam, largest harbour in the world, 10 million people living below or closely at sea-level. I hope you get the picture.
What was the Corps of Engineers thinking in the US? That a 1 in 200 years event wouldn't happen in their lifetime? Well, given an expected age of 80 years, 40% chance that is does. For a single event! This is a situation that was known and budgetted for for at least 40 years. And now the Americans are all in tears!
People, you have willfully neglected the risks for an event like this, probably because it's a cheap way to run a government. Now you are hurting. Please start to consider costs and benefits, examine the cost that would have incurred if the levees were brought up to par in the sixties, compare it with the costs now (excluding the unnecessary loss of life), and make sure it won't happen again.
Ah, who am I kidding. The US as a noob nation is incable of looking ahead more than a few years at a time. This in another fact that has been proven time and time again. When will you learn to consider life more important than the almighty buck.
Irony: God bless America.
If he's offering a box set next to a 65 hours/week job, he's a fucking idiot and deserves all he gets.
Joe Blow should know better than to run a software business next to his 65 hour work week. Oh, Open Source you say? You mean like the kind that has no warranty, no fitness for a particular purpose and no monetary value whatsoever being exchanged?. Don't worry, Joe Blow hasn't sold a thing to anyone, he is safe.
What do you think the shareholders use to track their investments?
A more apt comparison would be to say that IBM was a huge player in the mainframe market. And where are they now? They're still a huge player in the mainframe market. It is just that this market has slowly eroded, and IBM is slowly changing to accommodate this. Likewise, only when the PC market will dwindle does Microsoft have anything to fear. Finally, 40 billion dollars go a long way for living through hard times. Don't count on Microsoft going away in the next few decades.
Unfortunately it seems that slashdot these days is mostly populated by morally indignant, self-righteous jerks that lack any semblance of a sense of humour. The situation doesn't call for a heavy-handed set of arguments about the relative moralities of the parties involved. The situation is simply funny, laugh.
Still I think taking the toy away is the better course of action. Access restrictions on the routers might also be something to contemplate in such an environment, but maybe the schools really aren't up to such a technological challenge. But going through heavy duty legal stuff to make up for their own incompetence to handle their pupils doesn't really sit well with me.
Actually, we do try to keep the passwords in sync more or less. It usually takes around three guesses to find it. The point is still that we're a small outfit and we don't have financial accounts to protect. The first step in security is determining what needs protection. In our outfit, internal access is open, though external access is purely through a stored public key on the access point. The private keys are personal.
Mambo renames to webportal
firefox renamed to browser
openoffice renamed to:
word processor
spread sheet
etc.
linux (and(!)/or BSD) renamed to 'operating system'
This is going to be a ball!
Two words and a questionmark: felony charges?
Note that we're working on windows machines, where it takes a good half day to create a workable development environment if you log in as a new user. Re-using a co-workers environment thus saves hours. As for outside intrusions, we have a lock on the door and an alarm as well.
In all fairness, all we need is a finite improbability drive. It's easy to make it infinite from there.