But it is a statistical necessity if all you know accurately is there were more than a thousand results doing anything else is pretty pointless.
The only other approach that springs to mind for completeness testing is to take a selection of web pages, and see if any of them are missing from each engine. But I suspect finding a good selection of missing pages is a lot harder than tossing out popular results.
Done over a weekend - sure. Inaccurate and imprecise - sure. Puts the onus on Yahoo! to put up or shut up - sure.
This study isn't answering the question, it is asking Yahoo! to justify it's claim.
I certainly wish people would stop publishing in medical journals without a proper statistician analysing the figures. I'm not so worried about Slashdot;)
Interesting question is should we count hits, or unique clients. Most people seem to want to know the "unique client" figure, but when designing a webpage raw hits is probably the thing to use. In which case for every 1 Gecko rendering hit, I see 6.4 Internet MSIE hits. The numbers are based on the last 1 million hits to our servers, so I'd expect confidence intervals to be quite small, but I haven't calculated them.
I dare say that may be 8% firefox per unique client as per the article, but it is 12% of ALL hits. Our figures have been pretty stable for a while, which seems "odd" because I know a lot of people who have switched to Firefox (not enough to warp our figures, but the stats don't agree with what people are telling me.
I've seen Firefox misbehaving on Windows, but it does seem to be specific machines. Unfortunately due to the "per file" revision manangement in Windows, the best answer is "get a better OS", or reinstall and figure out what breaks it.
I've seen crashes on an old Redhat box, but I assume that is due to using the Firefox Linux distro rather than proper Redhat packages.
Where it is properly packaged, in a properly controlled environment it seems pretty solid.
"No, a web designer wants to design his website to hit the majority of his audience."
I'm sure he does. I assumed the original poster was referring to the fact that IE rendering is very buggy, so it is very hard to tell if your web page will render correctly in IE without testing every little change, where as almost any other browser will do what it says in the standards.
Indeed almost every page I write, I write to the standards, and then spend ages wondering why it messes up in IE. Realistically this is the only way to get 100% coverage. Standard + minimal IE workarounds (assuming you can find a standard compliant IE workaround).
Re:You can't catch it all
on
Ending Spam
·
· Score: 1
First someone needs to invent a methodology for communication that prevents spam.
No good saying SMTP is broke, if you don't say how to fix it conceptually. Otherwise the solution will end up as spam free as SPF is (SPF never set out to stop spam, but to stop impersonation, but that isn't how it was sold).
Ultimately the protocol has to have a way to shifting costs to the sender, or at least offer that option , and these costs may have to be larger than the cost of the communication itself (which may be neglible).
I personally believe the brokenness of SMTP is exagerrated, and the brokeness of the many client machines running botnets is underestimated.
The problem with the book, is it addresses how to do content filtering, when it isn't the content of spam that is the problem. The old adage about scientists making a living studying a problem, and engineers making a living avoiding such problems, springs to mind.
Which is why methods that address the underlying issues (bulk, unsolicited, compromised boxes) often work better (think whitelisting, greylisting, challenge/response (whatever you think of it, it is effective for the person using it), Exploit block lists) than statistical filtering, or statistical weighting of a number of rules.
"I just realized I suggested a 'gentleman's wager' on slashdot...to an AC...Doh!"
Hehe, although you caused me (not the AC) to check my electricity costs.
Seems I can buy off-peak electricity at about 0.06$/KwH, so for as long as I run Hydrogen generation during the night. I just need to go see how well the fuel cells do per cubic meter....
However I also saw the cost of Hydrogen generators, let us merely say they haven't become "consumer items" yet. But then at tens of cubic meters an hour, they are obviously designed for commercial use.
Although I suspect with a little help from my engineering guru friend, I can make a home sized hydrogen generator for a miniscule fraction of the retail price. I recall it being fairly easy to extract hydrogen from water, with electricity, when I was at school.
I was looking around at pure electric cars recently, just to see what is out there. And it does look like improvements in battery technology may finally be making it feasible to use these for more than short urban commutes. Although I want to see it working before I buy.
But then most days my car only does work and back, which the lead/acid electric cars would do fine.
My criteria though is my dad lives about 320 miles away, and anything that can get me there with only one moderately long stop for charging (I need recharging on 6 hour car journeys anyway, and am quite happy to spend 2+ hours over recharging me - purely if the car needs such a long stop you understand!).
I need to offset the cost of buying a TZero. Perhaps if I can find enough people at/. to offer me good odds on my next car not being electric.
The current diesel is in need of some welding, and the current "best electric car" appears to be a Reva (which no self respecting geek would be seen dead in - although might be seen trying to wire into the national grid halfway down the motorway to help a stranded motorist). So if you guys could see your way to offering me odds of about 100 to 1 against......
Netcraft exclude recent Linux kernels, because of the 47 days rollover in the method they use to calculate.
The uptime project tracks uptime for fun... Anyone have a feel for how valid they are?
Basically they have currently (ignoring OSes with minute share).
Linux 2.2 Best Linux 2.4 Linux 2.6 Windows 2000 Windows 2003 Worst
I'll ignore the XP stats, let us charitably assume a lot of the people didn't realise it was a competition;)
No doubt older OSes/kernels benefit from a few very big uptimes. But at 20d or so Windows 2003 is certainly looking "stable". Suggests some of the perceived convergence might be those damned recent Linux kernels. I think I may have a resource leak in netfiler:(
Sure it must be possible to correct these stats for the few very high uptimes - 90% range perhaps.
Where differences start to tell is when I patch my Debian boxes, and replace one major server app with another, the box and the rest of it's services just keep plodding along. Barely glancing at W2K3 will cause it to ask for a reboot still (I remember the W2K launch party where they claimed 26 cases where unnecessary reboots were required had been removed.... they didn't say how many were left).
"We laugh in the face of microsoft for using this argument"
I don't, statistics aren't that difficult to grasp.
We laugh at Microsoft for using this as an excuse to cover extremely poor design decisions, stupid choice of defaults, and desparately badly thought out remedies.
But wait Vista will have security built into it from the base or something, presumably implying that all those previous Microsoft OSes didn't. Which is terribly unfair on the NT team, who built it in, and had it engineered out on the grounds of backward compatibility.
I think you may be looking at ones that measure network performance, rather than specifically server performance or reliability.
There aren't any pure Windows solution in the netcraft longest uptime top 10, the first when I looked was in 26th place, II5 on W2K. There are some highbred solutions (IIS on BSD) presumably firewalled or proxies (we have IIS on Linux due to squid accelerator being used at work) further up the list.
Of course BSD dominates because they didn't have a roll-over in the reported uptime counter, unlike a certain OS from Finland.
Of course stats only tell half the story, whilst we had some kernel trouble with our Linux firewall, a reboot and a head scratch is nothing as to the mysteries that W2K3 has thrown up, including two known Microsoft bugs, for which there is no fix. Microsoft code written by people who don't understand the Microsoft user model (not surprising given how complex it has become under ADS).
No way would I willingly trust big Enterprise systems to W2K3, it just doesn't look ready to me. My employers small enterprise is depending on it, but then all it is doing is being a file and authentication server. If we had a Linux box doing that role, I'd be very surprised if we tripped over major bugs like we did with W2K3.
Netcraft suggests it will be about another 8 months before tommy.com can claim equivalent stability for the new OS, given it appears their GNU/Linux servers "just worked".
My guess is new management wanted to change things to something they feel more comfortable with. Seen that at a lot of places, it usually plays merry hell with the service availability stats.
Not fiddling is the key to good availability, and IT folk are nothing if not keen fiddlers. I fiddled today and broke stuff, and I know better.
My desktop experience is fine. But then my desktop boxes have both been up for longer than the tommy.com W2003 servers, and I value that in a desktop. They would have been up a lot longer if I had them both on UPSes .
10,000 thousand people migrate their web services to Linux isn't news, 1 person migrates web services from Linux to Windows is news.
"also most of the problems on windows are well known viruses. cleaning up what you belive is a deliberate attack on YOUR system would obviously justify far more care."
I thought the whole point of the article was that the common malware may be being used for uncommonly nefarious purpose. Just because 10,000 people got hit by the same malware doesn't make it any less specific a threat to you. The "My city got hit by a nuke, so it is okay as they weren't targeting me personally" logic.
People have to learn that as soon as someone finds a way to get malware on your box it is effectively game over. If one person does it undetected, so can someone else. Reinstall.
'The problem is that many of the "holes" have already been filled'
Actually this is the "god of the gaps" hypothesis. Everything we don't understand is done by god. So until we discovered microbes - illness was a curse from god.
The serious theologians have backed off this position, as they have seen science explain all sorts of things that were "god", and realised it is potentially an unsustainable position.
Of course you have to ask why an intelligent designer allows all those other unhelpful mutations to happen alongside his preferred ones? Presumably out of a sense of malice, or to maintain the illusion it is a process of chance by making it look indistinguishable from chance.
"it cannot be denied that Microsoft's development methods are demonstrably capable of producing quality software"
Which piece of software are you thinking of?
Sorry I know Excel is kind of okay, if you can put up with a spreadsheet that gives the wrong answers occaisonally. But in the business arena there is very little MS software I'd ever want to run on a computer.
Their principal development technique, has been build it quick and dirty, and undercut the competition, growing the business from the bottom end. This tends to produce qucik and dirty software that has been patched till it nearly works.
Excel, and Visio are kind of okay, but Word, Powerpoint, Access, any of the OSes (except perhaps W2K), IIS, IE, VSS, forget it they are worse than the free alternatives, let alone the proprietary alternatives.
"If I had the home address and phone number of someone that was unpopular on Slashdot should I have the right to post it? Should I have the right to lie about them?"
Ultimately yes, Freenet gives you that freedom, but the data will disappear very rapidly if no one reads it. Also the credibility of the information depends on the digital pseudonym under which it is "published" and "signed".
Why would anyone believe you if you said Bill Gates was at 211b Bakers Street?
Okay someone could publish a series of spammers/pedophiles and make a mistake. Anonymous publication changes the trust relationship between reader, and writer.
Kind of academic, freenet (or freenets - who knows if they are all connected?) has been around for ages, all that is announced is a version with different features (possibly more scalable), unless it is made illegal to run a node (and any other anonymous port based protocol... otherwise how do you tell) freenet is here.
I vaguely recall from the Falklands war, that the time between an incoming Exocet missile coming over the horizon, and hitting a ship, was of the order of tens of seconds.
Okay not all incoming hostile forces are doing MACH 3, at 15 foot above the ground, but that is "ancient" technology.
Comment was made (possibly untrue) that the computers automatically determined the incoming to have been French, and thus probably not hostile, and so didn't use the remaining seven seconds to try and shoot it out of the air before impact.
Haldemann extrapolated/anticipated this kind of thing in "The Forever War".
In such scenarios there is no time for meaningful human decision making, and you either work to avoid the scenario, or risk shooting down aircraft that fail to automatically identify themselves promptly (as happened to the Iranian airliner, or possibly the identification systems failed).
Putting humans in such a decision loop just means more delays and more mistakes.
There are some simple corporate tests as to whether a company is "worth it".
The simplest is replacement cost, given three billion USD (plus any revenues of course) could you create a new Skype?
Hard to assess the value of the userbase, and the telephone Interconnects, and goodwill, but given the immaturity of the market, and the speed with which Skype emerged, my gut feeling is yes I could replace it for less than 3Bn USD.
But then maybe there are aspects of the deal not immediately obvious, and simply because I think the price discussed is slightly inflated price doesn't mean the buyer can't make a profit. Let us call it "investment bank surplus".
"I still think that people are willing to pay for stability and security. If not, why are their more desktop machines running mac os then linux?"
The point is there are hardly any of either. But if you forced me to answer I'd say you've been able to buy MACs with OSX preinstalled much more easily.
Sure there are some people willing to pay for quality in any market.
But the bulk of the computer market appears to be price driven. I fear it is because the customers don't understand what they are buying, so don't know how to compare one to the other.
A good example is the MHz (of GHz these days) fallacy of processors, that processors get faster with increasing MHz numbers. Which may be true within one family of processors.
Similar effects were seen in the firewall market, where every vendor was keen to tell you how many Megabits per second they could ship. Rather than going head to head on say the number of protocols they could apply application level proxying to say, or other aspects of security.
Whilst I think the closed hardware argument carries some weight - I've seen MacOSX crash - given the amount of time I've seen OSX running that is not good!
When I see my Linux boxes crash I immediately start checking the hardware for failures, or just replace the hardware, haven't had that sort of confidence in OS software since I was HP-UX admin fulltime.
So it isn't clear to me that OSX is more stable than Linux (although the code OSX kernel is based on appears to be).
People don't need to hunt for drivers if you build and ship PCs with GNU/Linux preinstalled. This sort of OEM work is why Windows runs at all on seemingly random hardware. The model for making an an OS work on cheap commodity hardware is well established - although I sometimes wonder if DELL are forgetting how to do it.
I don't believe people are prepared to pay for security or stability in a desktop operating system. It may be insanity, but the visitor figures to our websites speak for themselves.
People are loathed to pay more, even where there is an obvious gain in quality (Apple hardware is always well made in my experience). Most normal people are also loathed to change, or fiddle with their computers.
As a result people may well adopt GNU/Linux over OSX, because it is cheaper, but they'll only do it when the Wallmart's of this world offer it preinstalled. Smaller companies can't do this, because the OEM work to make a successful and featured computer running GNU/Linux is expensive and needs to be spread over a large number of units to be competitive (with MS Windows). Indeed hardware purchases need to be spread over a large number of units to be price competitive these days.
One occaison when I recommended Apple laptops to a client, you'd have thought I'd suggested they try unnatural acts with a chicken. Even though it fitted their needs perfectly.
The market out there for computers isn't rational or well informed, but it is price conscious.
Re:Stop pandering to the lowest common denominator
on
Improving Education?
·
· Score: 1
"Teachers spend so much time trying to teach the dumb students"
Scarily your sentiments were concurred with in a recent UK report, that noted a number of heads had diverted funds specially earmarked for gifted students to other programmes.
I think it comes to the aforementioned anti-intellectual ethos, which seems to be at least Anglo-American in nature, but I'm told isn't universal across all cultures.
Slashdot Mirror
But it is a statistical necessity if all you know accurately is there were more than a thousand results doing anything else is pretty pointless.
The only other approach that springs to mind for completeness testing is to take a selection of web pages, and see if any of them are missing from each engine. But I suspect finding a good selection of missing pages is a lot harder than tossing out popular results.
Done over a weekend - sure.
Inaccurate and imprecise - sure.
Puts the onus on Yahoo! to put up or shut up - sure.
This study isn't answering the question, it is asking Yahoo! to justify it's claim.
"They're meaningless."
;)
I certainly wish people would stop publishing in medical journals without a proper statistician analysing the figures. I'm not so worried about Slashdot
Interesting question is should we count hits, or unique clients. Most people seem to want to know the "unique client" figure, but when designing a webpage raw hits is probably the thing to use. In which case for every 1 Gecko rendering hit, I see 6.4 Internet MSIE hits. The numbers are based on the last 1 million hits to our servers, so I'd expect confidence intervals to be quite small, but I haven't calculated them.
I dare say that may be 8% firefox per unique client as per the article, but it is 12% of ALL hits. Our figures have been pretty stable for a while, which seems "odd" because I know a lot of people who have switched to Firefox (not enough to warp our figures, but the stats don't agree with what people are telling me.
I've seen Firefox misbehaving on Windows, but it does seem to be specific machines. Unfortunately due to the "per file" revision manangement in Windows, the best answer is "get a better OS", or reinstall and figure out what breaks it.
I've seen crashes on an old Redhat box, but I assume that is due to using the Firefox Linux distro rather than proper Redhat packages.
Where it is properly packaged, in a properly controlled environment it seems pretty solid.
"No, a web designer wants to design his website to hit the majority of his audience."
I'm sure he does. I assumed the original poster was referring to the fact that IE rendering is very buggy, so it is very hard to tell if your web page will render correctly in IE without testing every little change, where as almost any other browser will do what it says in the standards.
Indeed almost every page I write, I write to the standards, and then spend ages wondering why it messes up in IE. Realistically this is the only way to get 100% coverage. Standard + minimal IE workarounds (assuming you can find a standard compliant IE workaround).
First someone needs to invent a methodology for communication that prevents spam.
No good saying SMTP is broke, if you don't say how to fix it conceptually. Otherwise the solution will end up as spam free as SPF is (SPF never set out to stop spam, but to stop impersonation, but that isn't how it was sold).
Ultimately the protocol has to have a way to shifting costs to the sender, or at least offer that option , and these costs may have to be larger than the cost of the communication itself (which may be neglible).
I personally believe the brokenness of SMTP is exagerrated, and the brokeness of the many client machines running botnets is underestimated.
The problem with the book, is it addresses how to do content filtering, when it isn't the content of spam that is the problem. The old adage about scientists making a living studying a problem, and engineers making a living avoiding such problems, springs to mind.
Which is why methods that address the underlying issues (bulk, unsolicited, compromised boxes) often work better (think whitelisting, greylisting, challenge/response (whatever you think of it, it is effective for the person using it), Exploit block lists) than statistical filtering, or statistical weighting of a number of rules.
Odds of 200 to 1, and I can probably pick up a Venturi Fetish... You know you want to.
"I just realized I suggested a 'gentleman's wager' on slashdot...to an AC...Doh!"
e .htm
/. to offer me good odds on my next car not being electric.
Hehe, although you caused me (not the AC) to check my electricity costs.
Seems I can buy off-peak electricity at about 0.06$/KwH, so for as long as I run Hydrogen generation during the night. I just need to go see how well the fuel cells do per cubic meter....
However I also saw the cost of Hydrogen generators, let us merely say they haven't become "consumer items" yet. But then at tens of cubic meters an hour, they are obviously designed for commercial use.
Although I suspect with a little help from my engineering guru friend, I can make a home sized hydrogen generator for a miniscule fraction of the retail price. I recall it being fairly easy to extract hydrogen from water, with electricity, when I was at school.
I was looking around at pure electric cars recently, just to see what is out there. And it does look like improvements in battery technology may finally be making it feasible to use these for more than short urban commutes. Although I want to see it working before I buy.
But then most days my car only does work and back, which the lead/acid electric cars would do fine.
My criteria though is my dad lives about 320 miles away, and anything that can get me there with only one moderately long stop for charging (I need recharging on 6 hour car journeys anyway, and am quite happy to spend 2+ hours over recharging me - purely if the car needs such a long stop you understand!).
http://www.acpropulsion.com/tzero_pages/tzero_hom
I need to offset the cost of buying a TZero. Perhaps if I can find enough people at
The current diesel is in need of some welding, and the current "best electric car" appears to be a Reva (which no self respecting geek would be seen dead in - although might be seen trying to wire into the national grid halfway down the motorway to help a stranded motorist). So if you guys could see your way to offering me odds of about 100 to 1 against......
Step 6 - Prophet
Netcraft exclude recent Linux kernels, because of the 47 days rollover in the method they use to calculate.
;)
:(
The uptime project tracks uptime for fun... Anyone have a feel for how valid they are?
Basically they have currently (ignoring OSes with minute share).
Linux 2.2 Best
Linux 2.4
Linux 2.6
Windows 2000
Windows 2003 Worst
I'll ignore the XP stats, let us charitably assume a lot of the people didn't realise it was a competition
No doubt older OSes/kernels benefit from a few very big uptimes. But at 20d or so Windows 2003 is certainly looking "stable". Suggests some of the perceived convergence might be those damned recent Linux kernels. I think I may have a resource leak in netfiler
Sure it must be possible to correct these stats for the few very high uptimes - 90% range perhaps.
Where differences start to tell is when I patch my Debian boxes, and replace one major server app with another, the box and the rest of it's services just keep plodding along. Barely glancing at W2K3 will cause it to ask for a reboot still (I remember the W2K launch party where they claimed 26 cases where unnecessary reboots were required had been removed.... they didn't say how many were left).
"We laugh in the face of microsoft for using this argument"
I don't, statistics aren't that difficult to grasp.
We laugh at Microsoft for using this as an excuse to cover extremely poor design decisions, stupid choice of defaults, and desparately badly thought out remedies.
But wait Vista will have security built into it from the base or something, presumably implying that all those previous Microsoft OSes didn't. Which is terribly unfair on the NT team, who built it in, and had it engineered out on the grounds of backward compatibility.
What netcraft stats you checking?
I think you may be looking at ones that measure network performance, rather than specifically server performance or reliability.
There aren't any pure Windows solution in the netcraft longest uptime top 10, the first when I looked was in 26th place, II5 on W2K. There are some highbred solutions (IIS on BSD) presumably firewalled or proxies (we have IIS on Linux due to squid accelerator being used at work) further up the list.
Of course BSD dominates because they didn't have a roll-over in the reported uptime counter, unlike a certain OS from Finland.
Of course stats only tell half the story, whilst we had some kernel trouble with our Linux firewall, a reboot and a head scratch is nothing as to the mysteries that W2K3 has thrown up, including two known Microsoft bugs, for which there is no fix. Microsoft code written by people who don't understand the Microsoft user model (not surprising given how complex it has become under ADS).
No way would I willingly trust big Enterprise systems to W2K3, it just doesn't look ready to me. My employers small enterprise is depending on it, but then all it is doing is being a file and authentication server. If we had a Linux box doing that role, I'd be very surprised if we tripped over major bugs like we did with W2K3.
"Am I missing something here?"
Difference between reality and what people say?
Netcraft suggests it will be about another 8 months before tommy.com can claim equivalent stability for the new OS, given it appears their GNU/Linux servers "just worked".
My guess is new management wanted to change things to something they feel more comfortable with. Seen that at a lot of places, it usually plays merry hell with the service availability stats.
Not fiddling is the key to good availability, and IT folk are nothing if not keen fiddlers. I fiddled today and broke stuff, and I know better.
My desktop experience is fine. But then my desktop boxes have both been up for longer than the tommy.com W2003 servers, and I value that in a desktop. They would have been up a lot longer if I had them both on UPSes .
10,000 thousand people migrate their web services to Linux isn't news, 1 person migrates web services from Linux to Windows is news.
If it was free they wouldn't have to purchase it either, so no press release.
Tomorrows story - Indiana discover 100K desktops already running Debian.
"also most of the problems on windows are well known viruses. cleaning up what you belive is a deliberate attack on YOUR system would obviously justify far more care."
I thought the whole point of the article was that the common malware may be being used for uncommonly nefarious purpose. Just because 10,000 people got hit by the same malware doesn't make it any less specific a threat to you. The "My city got hit by a nuke, so it is okay as they weren't targeting me personally" logic.
People have to learn that as soon as someone finds a way to get malware on your box it is effectively game over. If one person does it undetected, so can someone else. Reinstall.
'The problem is that many of the "holes" have already been filled'
Actually this is the "god of the gaps" hypothesis. Everything we don't understand is done by god. So until we discovered microbes - illness was a curse from god.
The serious theologians have backed off this position, as they have seen science explain all sorts of things that were "god", and realised it is potentially an unsustainable position.
Of course you have to ask why an intelligent designer allows all those other unhelpful mutations to happen alongside his preferred ones? Presumably out of a sense of malice, or to maintain the illusion it is a process of chance by making it look indistinguishable from chance.
I thought the SAMBA team had already given advice to the SMB protocol people, so surely it should be modded redundant.
"it cannot be denied that Microsoft's development methods are demonstrably capable of producing quality software"
Which piece of software are you thinking of?
Sorry I know Excel is kind of okay, if you can put up with a spreadsheet that gives the wrong answers occaisonally. But in the business arena there is very little MS software I'd ever want to run on a computer.
Their principal development technique, has been build it quick and dirty, and undercut the competition, growing the business from the bottom end. This tends to produce qucik and dirty software that has been patched till it nearly works.
Excel, and Visio are kind of okay, but Word, Powerpoint, Access, any of the OSes (except perhaps W2K), IIS, IE, VSS, forget it they are worse than the free alternatives, let alone the proprietary alternatives.
"If I had the home address and phone number of someone that was unpopular on Slashdot should I have the right to post it? Should I have the right to lie about them?"
Ultimately yes, Freenet gives you that freedom, but the data will disappear very rapidly if no one reads it. Also the credibility of the information depends on the digital pseudonym under which it is "published" and "signed".
Why would anyone believe you if you said Bill Gates was at 211b Bakers Street?
Okay someone could publish a series of spammers/pedophiles and make a mistake. Anonymous publication changes the trust relationship between reader, and writer.
Kind of academic, freenet (or freenets - who knows if they are all connected?) has been around for ages, all that is announced is a version with different features (possibly more scalable), unless it is made illegal to run a node (and any other anonymous port based protocol... otherwise how do you tell) freenet is here.
I vaguely recall from the Falklands war, that the time between an incoming Exocet missile coming over the horizon, and hitting a ship, was of the order of tens of seconds.
Okay not all incoming hostile forces are doing MACH 3, at 15 foot above the ground, but that is "ancient" technology.
Comment was made (possibly untrue) that the computers automatically determined the incoming to have been French, and thus probably not hostile, and so didn't use the remaining seven seconds to try and shoot it out of the air before impact.
Haldemann extrapolated/anticipated this kind of thing in "The Forever War".
In such scenarios there is no time for meaningful human decision making, and you either work to avoid the scenario, or risk shooting down aircraft that fail to automatically identify themselves promptly (as happened to the Iranian airliner, or possibly the identification systems failed).
Putting humans in such a decision loop just means more delays and more mistakes.
There are some simple corporate tests as to whether a company is "worth it".
The simplest is replacement cost, given three billion USD (plus any revenues of course) could you create a new Skype?
Hard to assess the value of the userbase, and the telephone Interconnects, and goodwill, but given the immaturity of the market, and the speed with which Skype emerged, my gut feeling is yes I could replace it for less than 3Bn USD.
But then maybe there are aspects of the deal not immediately obvious, and simply because I think the price discussed is slightly inflated price doesn't mean the buyer can't make a profit. Let us call it "investment bank surplus".
"on the grounds that the IS team had not done a security review of FF, but they had of IE."
You have to realise at this point that;
a) they are bullshitting you.
or
b) the IS team is incompetent since they did a security review on IE *AND* still approved it.
About that time you should have started looking for the next contract.
In another window is a discussion of DABS selling PCs with Mandrake for 140GBP (excluding VAT ~ 246 USD plus taxes).
Although I think possibly DABS isn't set to provide this kind of product, perhaps there is a service company behind it?
"I still think that people are willing to pay for stability and security. If not, why are their more desktop machines running mac os then linux?"
The point is there are hardly any of either. But if you forced me to answer I'd say you've been able to buy MACs with OSX preinstalled much more easily.
Sure there are some people willing to pay for quality in any market.
But the bulk of the computer market appears to be price driven. I fear it is because the customers don't understand what they are buying, so don't know how to compare one to the other.
A good example is the MHz (of GHz these days) fallacy of processors, that processors get faster with increasing MHz numbers. Which may be true within one family of processors.
Similar effects were seen in the firewall market, where every vendor was keen to tell you how many Megabits per second they could ship. Rather than going head to head on say the number of protocols they could apply application level proxying to say, or other aspects of security.
Whilst I think the closed hardware argument carries some weight - I've seen MacOSX crash - given the amount of time I've seen OSX running that is not good!
When I see my Linux boxes crash I immediately start checking the hardware for failures, or just replace the hardware, haven't had that sort of confidence in OS software since I was HP-UX admin fulltime.
So it isn't clear to me that OSX is more stable than Linux (although the code OSX kernel is based on appears to be).
People don't need to hunt for drivers if you build and ship PCs with GNU/Linux preinstalled. This sort of OEM work is why Windows runs at all on seemingly random hardware. The model for making an an OS work on cheap commodity hardware is well established - although I sometimes wonder if DELL are forgetting how to do it.
I don't believe people are prepared to pay for security or stability in a desktop operating system. It may be insanity, but the visitor figures to our websites speak for themselves.
People are loathed to pay more, even where there is an obvious gain in quality (Apple hardware is always well made in my experience). Most normal people are also loathed to change, or fiddle with their computers.
As a result people may well adopt GNU/Linux over OSX, because it is cheaper, but they'll only do it when the Wallmart's of this world offer it preinstalled. Smaller companies can't do this, because the OEM work to make a successful and featured computer running GNU/Linux is expensive and needs to be spread over a large number of units to be competitive (with MS Windows). Indeed hardware purchases need to be spread over a large number of units to be price competitive these days.
One occaison when I recommended Apple laptops to a client, you'd have thought I'd suggested they try unnatural acts with a chicken. Even though it fitted their needs perfectly.
The market out there for computers isn't rational or well informed, but it is price conscious.
"Teachers spend so much time trying to teach the dumb students"
Scarily your sentiments were concurred with in a recent UK report, that noted a number of heads had diverted funds specially earmarked for gifted students to other programmes.
I think it comes to the aforementioned anti-intellectual ethos, which seems to be at least Anglo-American in nature, but I'm told isn't universal across all cultures.