Most commercial projects of 300+ people fail to complete. The failure rate goes up exponentially as the size increases.
Very few 300 person projects would take only 6 months - you just cannot ramp up that quickly and maintain any sort of coherence.
A good example would be MSFT Word for Windows 1.0. Took over 5 years but the plan never said it was more than 1 year from completion. ("Rapid Development" by Steve McConnell has the details).
Most commercial products are abandonded by their vendors. A key decision factor when buying a product is - will this vendor be around in 12 months? The idea that in buying a commercial product you are buying an assurance of future upgrades would be laughable were it not so serious.
Even within one company I have had the same company offer me three different and incompatible products in three successive years for the same problem.
As well as the fixes to the IA architecture mentioned, they have also fixed the floating point register stack. It is very hard to write fast floating point for IA32 and the new architecture makes it easy.
Intel has had plenty of over-engineered disasters before (32432, 286) and Merced is another one IMO.
Ada as a feminist icon. According to an article in Scientific American a while back the reason Ada Lovelace became a math geek was that her mother did not want her having anything to do with her father Lord Byron the poet. So she raised her as a science person.
Ada is also famous as the person who wrote the first computer program known to have had a bug in it. It had several actually.
More seriously if the article linked to above is one of the best, then there is precious so show the low numbers of women in science are due to discrimination.
The article's approach is to 1) Start each article with a straw man quote from some sexist buffoon. 2) Try and muddy the waters about the research on differences in cognition and preferences between the sexes c) Give lots of anecdotal evidence of people's attitudes.
This is not the way to make a rational case for your position (unless you're a feminist academic where this sort of thing goes down well due to a lack of rigour in this arena).
I used to believe this stuff until I had a daughter. Parents give girls dolls and boys gns because that is what they want.
One thing to bear in mind, that confounds the sexist bigots though is that people are individuals and some women are interested in and good at science and some men are good at and interested in verbal things. So you need to treat everyone as an individual.
You will not get the stats desired by feminists without coercion though (eg the Soviet Union).
Stock market crash should interest slashdotters
on
Tech Stocks Tumble
·
· Score: 1
Why?
1. The stock market is the best, most forward looking, leading indicator of where the ecionomy is going. If the market crashes, it is fairly likely that the economy will be in the dirt in 12-24 months. That affects everyone, because the economy drives the job market.
2. High stock prices lower the cost of funds to companies. Thus you get companies like Redhat raising large sums of money and they are using a lot of that money to improve Linux, GCC etc. If that money dries up - which is likely after a major crash - life will be tougher for Linux. Some of those who work for those companies may be looking for work again.
- argument by assertion. It is a problem because I say so. It must be due tro discrimination because I say so.
Second:
- blatant derogatory generalisations about boys (they are "aggressive" and scare the girls away from the computers). I find this offensive and sexist.
Third
- still doesn't accept that there are biologically mediated differences between men's and women's brains (which do affect different people in different degrees, so yes some girls are good at maths). We have been hearing this nonsense since the 60s. It isn't true, sister. Open your eyes. And your mind.
Would you - could you - fly in a plane built using a "feminist physics" or a feminist engineering?
Generally you can expect a bug per 100 to 1000 lines of code written by wage slaves/Dilberts. So with 20,000,000 LOC, you are looking at 20,000 to 200,000 bugs.
Every Microsoft product I have used has had bugs that have inconvenienced me. But they don't do this by accident. They would rather release function with bugs sooner than release later with higher quality. Mostly it doesn't hurt them. 'Worse is better' because worse gets to market sooner.
This strategy does not work in the enterprise critical server market, because downtime is unacceptable. It will be difficult for Microsoft to make the culture change required to really compete in this arena. 20 years of 'near enough's good enough' doesn't change overnight.
I will not be impressed until I see the GCC patches - showing how well they have been able to optimize for the IA64; and until I see some benchmarks. The fact that Intel has had silicon for a while but no benchmarks so my knowledge tells me to suspect that it is not actually that fast.
Intel has had plenty of disasters in the past (the X87 stach architecture, the 16-bit segments, the 432 CPU).
AMD's web page has a good critique of IA64 that definitely puts the seeds of doubt in my mind see http://www.amd.com/products/cpg/mpf/pres99/micropf orum.html
> As well the IA64 architecture is *awesome*. 128 64-bit general purpose registers, an additional 128 64-bit floating point registers, and much much more. The coding that I am doing runs like 10x faster on a 666MHz IA64 than it does on a 800MHz PIII (literally!).
I would be interested to know the kind of app this is. Also whether you are coding in assembler or what other language.
I am having trouble seeing where the 10x performance would come from for most applications.
I have been using XP for a while, and I found that the whole is more than the sum of the parts. True, no element is entirely new, but when you put them together everything works.
Pair programming is an example. The previous poster claimed without any evidence that programming alone is more productive. I don't think it is, when you take into account that with pairs you get fewer errors, cleaner code, and better distribution of knowledge across the team. And less need for doco.
The existence of unit tests for example enables you to add features and clean up code later on. One reason for 'big design up front' is that the cost of changes later on is so high. But with the tests, you don't have to fear making changes. You can then defer adding features until you need them - which may be never. Thus unit testing enables 'just in time' coding.
I assume there is a huge witch hunt going on amongst those with big dollars at stake here.
Given that the crypto was only 40 bit, it was inevitable it would be cracked. Sooner rather than later. Why was it 40 bit. I assume this is due to the US government's export laws which are designed to ensure that the NSA can spy on whomever they please.
These laws are costing American companies a lot of money!
The other lesson from this is the lameness of almost all commercial security software. Security by obscurity does not work.
I friend of mine had the same experience. Sore wrists were getting to be a problem. Switched to dvorak and the problem went away.
I thought the article was very biased towards showing there is no problem with the free market that produced qwerty, and therefore is of little value.
I agree the research looked shoddy, but most research is flawed in some way. A few years back I researched heart disease and cholesterol. Most research was useless, due to lack of controls, too small a number of subjects, confounding variables, bias, failure to probe the full range of variation or to check all the variables, etc.
The complaints about flavours of Linux are just a rationalisation - the problems are of Sun's own making.
I have no problem writing code that runs on all the Linuxes - but I wrote a moderately large Java application and I had all sorts of compatibility problems between the JVMs. Not to mention JVM crashes and Win95 crashes, memory leaks, bugs in garbage collection etc. Back to C for me.
And it is obvious why the Linux JVM is not getting the support it needs - noone wants to sign away their rights to a *not* open source effort (no, not 'almost open' just not open).
Cryponomicon is full of errors. Maybe they are delierate but if so they seem pointless. Goto Dengo using the pole star to swim to the coast of New Guinea (south of the equator). RSA encryption's strength depends on the difficulty of factoring prime numbers etc.
Interestingly Bill Gates makes the same mistake about factoring prime numbers in his book 'The Road Ahead'. What a legend! (RSA depends on the difficulty of factoring composite numbers).
I have an old copy of Byte Magazine from 1997. The headline on the cover is "NT5 - ready for the enterprise!" This is what I used to call 'the Marketing present tense'. Referring to vapourware as if it exists.
In my experience when organisations get religion about documentation, they tend to produce lots of useless documents. The key documentation to produce is the maintenance manual and a users' guide. Keep both short and up to date.
Historical documents are useless eg those produced by various project phases, except for butt-covering in bureaucracies.
At the program level, document the interfaces to the programs and the expectations and assumptions, again quite concisely.
Some great reading on development is at c2.com/cgi/wiki?ExtremeProgrammingRoadmap.
This is a recurring pattern. Brave ideas about a reasonable crypto regime, them a major jellyback act as congress caves in.
Back in the J. Edgar Hoover days, when the FBI's budget was threatened, certain incriminating photos and other information would be brought to the attention of those involved.
Suddenly the FBI's budget would be off the agenda.
Anyone got a better theory about what happens with the crypto laws?
The article makes it pretty clear the aim of restricting crypto export is to make it easier to spy on people. The US has a long record of spying on friend and foe alike and using the information to protect its interests, especially commercial interests.
The fact that the export regulations make it considerably more difficult and inconvenient to get strong crypto WITHIN the US and presumably makes it easier to spy on US citizens - in the hypothetical scanario that they wanted to do that - is just a bonus.
Slashdot Mirror
On the point that cathedral projects have focus:
Most commercial projects of 300+ people fail to complete. The failure rate goes up exponentially as the size increases.
Very few 300 person projects would take only 6 months - you just cannot ramp up that quickly and maintain any sort of coherence.
A good example would be MSFT Word for Windows 1.0. Took over 5 years but the plan never said it was more than 1 year from completion. ("Rapid Development" by Steve McConnell has the details).
Most commercial products are abandonded by their vendors. A key decision factor when buying a product is - will this vendor be around in 12 months? The idea that in buying a commercial product you are buying an assurance of future upgrades would be laughable were it not so serious.
Even within one company I have had the same company offer me three different and incompatible products in three successive years for the same problem.
As well as the fixes to the IA architecture mentioned, they have also fixed the floating point register stack. It is very hard to write fast floating point for IA32 and the new architecture makes it easy.
Intel has had plenty of over-engineered disasters before (32432, 286) and Merced is another one IMO.
I saw the other day someone claimed the NSA can read PGP encrypted emails with ease. Should we bother trying to encrypt out emails?
OK tte NSA is not going to answer this but what do others think?
What about this? Leaving aside the NSA, can other people read PGP etc?
Ada as a feminist icon. According to an article in Scientific American a while back the reason Ada Lovelace became a math geek was that her mother did not want her having anything to do with her father Lord Byron the poet. So she raised her as a science person.
Ada is also famous as the person who wrote the first computer program known to have had a bug in it. It had several actually.
More seriously if the article linked to above is one of the best, then there is precious so show the low numbers of women in science are due to discrimination.
The article's approach is to 1) Start each article with a straw man quote from some sexist buffoon. 2) Try and muddy the waters about the research on differences in cognition and preferences between the sexes c) Give lots of anecdotal evidence of people's attitudes.
This is not the way to make a rational case for your position (unless you're a feminist academic where this sort of thing goes down well due to a lack of rigour in this arena).
I used to believe this stuff until I had a daughter. Parents give girls dolls and boys gns because that is what they want.
One thing to bear in mind, that confounds the sexist bigots though is that people are individuals and some women are interested in and good at science and some men are good at and interested in verbal things. So you need to treat everyone as an individual.
You will not get the stats desired by feminists without coercion though (eg the Soviet Union).
Why?
1. The stock market is the best, most forward looking, leading indicator of where the ecionomy is going. If the market crashes, it is fairly likely that the economy will be in the dirt in 12-24 months. That affects everyone, because the economy drives the job market.
2. High stock prices lower the cost of funds to companies. Thus you get companies like Redhat raising large sums of money and they are using a lot of that money to improve Linux, GCC etc. If that money dries up - which is likely after a major crash - life will be tougher for Linux. Some of those who work for those companies may be looking for work again.
All in all, definitely of interest I would think.
First:
- argument by assertion. It is a problem because I say so. It must be due tro discrimination because I say so.
Second:
- blatant derogatory generalisations about boys (they are "aggressive" and scare the girls away from the computers). I find this offensive and sexist.
Third
- still doesn't accept that there are biologically mediated differences between men's and women's brains (which do affect different people in different degrees, so yes some girls are good at maths). We have been hearing this nonsense since the 60s. It isn't true, sister. Open your eyes. And your mind.
Would you - could you - fly in a plane built using a "feminist physics" or a feminist engineering?
Generally you can expect a bug per 100 to 1000 lines of code written by wage slaves/Dilberts. So with 20,000,000 LOC, you are looking at 20,000 to 200,000 bugs.
Every Microsoft product I have used has had bugs that have inconvenienced me. But they don't do this by accident. They would rather release function with bugs sooner than release later with higher quality. Mostly it doesn't hurt them. 'Worse is better' because worse gets to market sooner.
This strategy does not work in the enterprise critical server market, because downtime is unacceptable. It will be difficult for Microsoft to make the culture change required to really compete in this arena. 20 years of 'near enough's good enough' doesn't change overnight.
I will not be impressed until I see the GCC patches - showing how well they have been able to optimize for the IA64; and until I see some benchmarks. The fact that Intel has had silicon for a while but no benchmarks so my knowledge tells me to suspect that it is not actually that fast.
f orum.html
Intel has had plenty of disasters in the past (the X87 stach architecture, the 16-bit segments, the 432 CPU).
AMD's web page has a good critique of IA64 that definitely puts the seeds of doubt in my mind see http://www.amd.com/products/cpg/mpf/pres99/microp
> As well the IA64 architecture is *awesome*. 128 64-bit general purpose registers, an additional 128 64-bit floating point registers, and much much more. The coding that I am doing runs like 10x faster on a 666MHz IA64 than it does on a 800MHz PIII (literally!).
I would be interested to know the kind of app this is. Also whether you are coding in assembler or what other language.
I am having trouble seeing where the 10x performance would come from for most applications.
The licence will only bind the parties to it.
Otherwise the general law of negligence applies.
So, for example, if my program to control robots goes beserk and kills a child who did not themselves have a licence, there is no protection.
Doctor Evil, one million dollars is not much these days. You should ask for ... TEN BILLION DOLLARS!!!
I have been using XP for a while, and I found that the whole is more than the sum of the parts. True, no element is entirely new, but when you put them together everything works.
Pair programming is an example. The previous poster claimed without any evidence that programming alone is more productive. I don't think it is, when you take into account that with pairs you get fewer errors, cleaner code, and better distribution of knowledge across the team. And less need for doco.
The existence of unit tests for example enables you to add features and clean up code later on. One reason for 'big design up front' is that the cost of changes later on is so high. But with the tests, you don't have to fear making changes. You can then defer adding features until you need them - which may be never. Thus unit testing enables 'just in time' coding.
Look at XP as a whole before judging it.
I assume there is a huge witch hunt going on amongst those with big dollars at stake here.
Given that the crypto was only 40 bit, it was inevitable it would be cracked. Sooner rather than later. Why was it 40 bit. I assume this is due to the US government's export laws which are designed to ensure that the NSA can spy on whomever they please.
These laws are costing American companies a lot of money!
The other lesson from this is the lameness of almost all commercial security software. Security by obscurity does not work.
I friend of mine had the same experience. Sore wrists were getting to be a problem. Switched to dvorak and the problem went away.
I thought the article was very biased towards showing there is no problem with the free market that produced qwerty, and therefore is of little value.
I agree the research looked shoddy, but most research is flawed in some way. A few years back I researched heart disease and cholesterol. Most research was useless, due to lack of controls, too small a number of subjects, confounding variables, bias, failure to probe the full range of variation or to check all the variables, etc.
The complaints about flavours of Linux are just a rationalisation - the problems are of Sun's own making.
I have no problem writing code that runs on all the Linuxes - but I wrote a moderately large Java application and I had all sorts of compatibility problems between the JVMs. Not to mention JVM crashes and Win95 crashes, memory leaks, bugs in garbage collection etc. Back to C for me.
And it is obvious why the Linux JVM is not getting the support it needs - noone wants to sign away their rights to a *not* open source effort (no, not 'almost open' just not open).
Cryponomicon is full of errors. Maybe they are delierate but if so they seem pointless. Goto Dengo using the pole star to swim to the coast of New Guinea (south of the equator). RSA encryption's strength depends on the difficulty of factoring prime numbers etc.
Interestingly Bill Gates makes the same mistake about factoring prime numbers in his book 'The Road Ahead'. What a legend! (RSA depends on the difficulty of factoring composite numbers).
I have an old copy of Byte Magazine from 1997. The headline on the cover is "NT5 - ready for the enterprise!" This is what I used to call 'the Marketing present tense'. Referring to vapourware as if it exists.
In my experience when organisations get religion about documentation, they tend to produce lots of useless documents. The key documentation to produce is the maintenance manual and a users' guide. Keep both short and up to date.
Historical documents are useless eg those produced by various project phases, except for butt-covering in bureaucracies.
At the program level, document the interfaces to the programs and the expectations and assumptions, again quite concisely.
Some great reading on development is at c2.com/cgi/wiki?ExtremeProgrammingRoadmap.
This is a recurring pattern. Brave ideas about a reasonable crypto regime, them a major jellyback act as congress caves in.
Back in the J. Edgar Hoover days, when the FBI's budget was threatened, certain incriminating photos and other information would be brought to the attention of those involved.
Suddenly the FBI's budget would be off the agenda.
Anyone got a better theory about what happens with the crypto laws?
The article makes it pretty clear the aim of restricting crypto export is to make it easier to spy on people. The US has a long record of spying on friend and foe alike and using the information to protect its interests, especially commercial interests.
The fact that the export regulations make it considerably more difficult and inconvenient to get strong crypto WITHIN the US and presumably makes it easier to spy on US citizens - in the hypothetical scanario that they wanted to do that - is just a bonus.