Complete and utter bollockssssss.... Fiasco of the Don Capellas cretinomoron - yes. Carly - no. Whatever people say about her any Compaq/HP shop can clearly tell you that during her reign the quality of hardware shipped drastically improved. If you have bought HPaq hardware over the last 6+ years you can clearly see the day when the cost saving idiot Capellas reign started. The quality of build immediately dropped to rock bottom. Passive cooling was replaced by cheap mexican fans which failed in 3-5 months, spec of the average box in fact dropped instead of rising. 512MB memory limit on a desktop down from 784 and thermal throttle on the CPU which effectively made it slower then the older models.
After that, during Carly it recovered to a more or less sane state - custom designed motherboards matched to the case instead of cheap chinese OEM shit, passive cooling assisted by the main chassis fan (an old Compaq TM), etc.
Whatever you say while I am not a Carly fan, she actually fixed a lot of the shit created by Don Cappellas. It would have been nice to see her go to be replaced by someone to improve further. Instead, she is being replaced by a Don Cappelas clone which means that all of us who buy Compaq are likely to have to look for an alternative vendor once again.
If it is worth working. Every time I talk to an agent jobhunting for them he is trying to offer me a 20% salary cut. To add insult to injury they also have the nerve to ask if your current salary is "negotiable". I have started answering "Yes, if you would like to negotiate it in the right direction, in other words - UP".
No thanks. They may have a few really smart people like Delany on their staff, but with this rate of pay I somehow doubt that they are going to get anywhere near getting and retaining talent in general
I know what I wrote. And I am a chemist by the way.
Water is a very complex substance. You have both van-der-vaals (weak) and polar (strong) interactions between the water molecules and dissolved substances which tend to be much more organised and complex then most average lamerz think. These can be altered by many factors including magnetic field and ionising radiation. After being altered they stay altered for days and sometimes weeks.
Realistically there are bottlenecks all over the place and out of them these 2 prevent nearly any computer from reaching 1G.
1. Interrupt handling bottleneck. Even with interrupt mitigation your typical pps value for a single CPU P4 is under 100 kps. It falls down to under 60 kps when using Intel dual CPUs (dunno about AMD or Via) or SMT due to the overly deep pipeline on the P4. That is way less then 1G for small packets.
2. IO bottleneck. Many motherboards have IO-to-memory speeds which are realistically way under 1G in total, usually around 600-700 Mbit.
What? The bomb detonated pretty much right between the two principal targets in the city, both Mitsubishi armaments factories.
And its target being the Mitsubishi naval yard which was 5 miles off. To add to that it also landed smack on top of a residential district with more then 50% of the ground 0 being a civilian target.
It was a war crime. Same as Dresden and same as Hiroshima.
How the f*** could it be selective and proper if the two idiots who dropped it misread the landmarks and dropped it more then 2 miles off target. It was in fact intended to hit the naval yards 5 km downstream from where it was dropped.
All Bulgarian, Cheh and most Caucasus (Russian and Georgian) SPAs are like this.
Radon containing water works miracles on arthritis, joints problems as well as many forms of eczema. While it usually fails to provide permanent cure it provides 3-4 months of relief or gives medications a better chance to work.
In btw, the feeling is weird... 20 minutes in a warm pool of such water makes you feel like your joints have started to melt. They feel like rubber.
The mechanism is still unclear, but it is not the Radon which is the active agent. It is the way its decay changes water properties.
If you are referring to the fact that Miller used to call his machines with different obsceneties in Russian like pizda (vagina), huy (dick), so on, I am well aware of that. None the less, sparc tree remains indicative of "quirky" hardware compared to ppc even after Millerisms have been cleaned out.
While both of them improve, Jo average speed of typing and speed of perception does not. As a result while the amount of data grows (flash, animations, video), the amount of items remains relatively constant (or grows at a much slowlier rate). Do not forget that the DOJ (or its equivalent elsewhere) can subpoena the data from the source or destination or both. Hence all it needs to see at the ISP level is that the data has been exchanged. Similarly, the fact that the data has been exchanged is sufficient to subpoena the content (Carnivore anyone?).
There is plenty of technology to do this now. No need for storage improvement. They can get it now and they are likely to get it.
As far as the kernel is concerned the number of profanities is a clear reflection of the quality of the underlying hardware. One of the things I do before buying new hardware is look at the comments in the linux kernel code. If they are like the ones you meet in the sun** architecture bit (it is the most profane part of the linux kernel) it may be a good idea to stay away.
For example just read the sunhme.c under drives/net. It is an absolute ROFL. Or arch/sparc/mm/ptrace.c...
You may be right as Sun has payed SCO to use their crap, but not to rerelease it under another license (and OSS to boot).
As anyone who has had to suffer from admin-ing OpenServer the idea of parallel non-init.d style startup launching is not new. In fact it is very old.
SCO had it first.
Before touching this "open" thing I will spend at least several days comparing this bit to the old OpenServer bit. And if they have as much in common as I suspect the best thing to do here will be to duck and wait for SCO legal missiles to land on a legitimate target.
In btw, that will be a nice way for McNealy to pay a bit more to SCO to top up his previous anti-linux contribution without causing undue screams in the community. All legit. Stolen code. Out of court settlement. Rather large one...
As a matter of fact, dear coward, my grandfather was a member of the russian group before he volunteered (it took him more then 5 tries to be reassigned) to follow his second profession and fight in the air and die as test pilot in the winter of 1942-43. If you are already that proficient in WW2 and Russian history you may have already guessed who (beware names are different). So, coming back to what is essentially a family history for me, the Russian effort was not espionage. At least not to a the extent traditionally assigned to it. Russian technology for graphite purification and reactor design were different. Bomb itself - yes, it was too similar to the American designs for everyone to become very suspicious. Getting to it and especially getting a working reactor - no.
There is something many WW2 historians have noticed.
Heisenberg was not an engineer and he did not get along with engineers. Typical of most European academentia at the time (and even now) . There were very few people to actually take the "ideas" and make them "tick" on the German nuclear team.
This is in sharp contrast with both American and Russian efforts which were done in a much more practical manner.
But the devil has been rumoured to drive a snowplough to work. Apparently he is fixing bugs in autofs, mailman, hpa-tftp(d) [which in fact is a linux kernel UDP bug]. Add glibc compat module to this.
To name just a few (from my "what do I follow and have tried to fix locally" list).
Well, besides that the penetration of computers into more and more parts of our life is decreasing the qualification of an average IT professional. Many people do IT without understanding anything about what they do. As a result the vendor services model is bound not to decrease. It is bound to grow.
In btw, when I saw the first forbes article I bet that it was PR placement by LMV. The second article proves it.
Real geeks have IJKL (and to lesser degree UOBN) worn out. AZ follow... But W, S, D - who cares about what weapon you are wielding, searching is best done via an artefact or intrinsic and D... WTF was D anyway...
Hmm... Circa 1997 80%+ of country MSFT ops ran their webservers on Linux or Solaris. The moment Netcraft published this and they became a laughing stock it was all migrated to IIS within 2 weeks. Similarly, I do not really believe in such posturing. MSFT is a marketing driven organisation and if their marketing decides that a specific instance of running Linux is bad for the current marketing campaigns there will be a big crater in its place in less then 5 seconds.
//begin sarcasm Or tactical planning simulator. How to prepare to blow your city council building in the most efficient manner. As a matter of fact it will be great if someone writes a map converter to load these into a decent flight simulator or into one of the first person shooters. And once again, quite usefull on planning your revenge on that pesky council clerk that did not give you planning permission for that nice cannon tower you were planning in your backyard//end sarcasm
On a more serious note, I somhow have the suspicion that some data will be unavailable or deliberately distorted. I simply do not see MSN publishing 45 degree 1m resolution pictures of Vandenberg, Clarke Fields or anything similar.
Dependencies? In a debian upgrade? Linux is not R00tH4t and R00tH4t aint linux. I think you need to run apt-get install -y lib-get-a-clue (with all dependecies)
And this is the exact reason why any good SLQ based system must have means of integrity checking.
As someone who have been writing database stuff for 10+ years now, I get really pissed off when I see lunatics raving on Acid about ACID. ACID in itself is not enough.
You must have reference checking, offline integrity tests as well as ongoing online integrity test. Repeating your example a transaction for buying tickets for a holiday must insert a record in the Requests table, Tickets table, Holidays table, etc and you must have an offline tool (even better backround thread) which checks that all records are present. In fact for the same reason in a well designed system you must violate 3rd normal form and have the integrity checking tool use the redundant data as well. Another alternative is a state load and checksum across the state storing it back in at least two different places (once again breaking 3rd normal form).
If you do it this way you can get a working system even if ACID breaks (databases have bugs), you can recover if hardware breaks and most importantly you have a considerable level of fraud resistance.
I have had so far 0 failure rate with them in servers with good cooling where they can be chilled down nicely to sub-30C due to the low profile which greatly improves airflow (knocking on wood here).
At the same time I have had 50%+ 1st year failure rate in Cubid mini ITX cases, 30%+ in Compaqs and other similar systems without dedicated disk fans. If their SMART sensors are to be believed they heat up to 42C+ in a HPaq and 50C+ in an ITX case.
If you run them in a desktop you must have all power management features enabled and must spindown the drive when not in use. If you do that, it will survive even in an mini ITX case. Otherwise your data will be toast. Literally.
Cisco has always been a champion in managing to ship a record number of bugs and still stay alive. Compare to them Microsoft is a stellar example of quality assurance.
Just look at the mixture of puke and shite they have for a TCP stack (do not mistake it with the forwarding engine which is a different thing). Every single TCP weakness reported in the last 3 years has cisco amidst the vulnerable list. Inadequate reaction to ICMPs, insecure resets as a reaction to bogus data, you name it.
Interestingly enough they are still out there despite all their bugs. My personal take on this is that they have managed to create create a stable ecosystem of certified "specialists" that do not know anything but cisco and do not buy anything but cisco. From there on it is not really relevant if it breaks.
Anyway, back on topic. Even a record number of bugs does not mean that a system is a usefull target for intrusion. Cisco is not such a target, because it is monolytic. You cannot easily transfer control from a bug to something that is usefull for you (some form of command line). There is nothing to "exec" so you have to figure out the address of an interpreter instance with enable access and manage to invoke it without buggering up the scheduler and the nearly insane memory allocation. Definitely non-trivial. Compare this to a unix platform where for a "good bug" you can execute a shell with under 30 bytes and the difference becomes quite obvious. Futher to that the CPU on an average cisco is outright flimsy so even after taking over one you cannot really use it to listen to network traffic. It has nearly no local storage so you cannot use it to stage an attack or as a zombie.
So on so fourth.
It is hard to hack and very useless. Once upon a time the answer to "I have root on this router, what can I do with it" was "You can shovel it up your arse". That is still the case.
In btw, I read the article and it does not compute. AFAIK Cisco (as a company) uses one-time passwords on all internet accessible systems. SecureID to be most exact. Used to be SKEY prior to 1998 (or 7?). So you can sniff passwords until you puke without getting anywhere. The article is missing a few steps on how the guy got from the bastion hosts to real stuff. Alternatively (and more likely) it is full of shit.
This means that you have not dealt with Telewest, Homechoice, Bulldog or any of the other "alternatives".
First, lets be clear. BT are in this for the money, not for the coolness. They have no intention on offering cool products dot-bomb style that do not bring profits.
Second, they may seem technologically backwater, but they are obliged by the UK regulatory regime to offer their products on a national basis. As a result if it takes to limit DSL to 512 (old Fujitsu linecards) or 1M (new linecards) to offer it for the coverage defined in the approved wholesale product they will do so without any regrets.
If you want BT to offer better products they happily will, provided that they can charge you for it and that the product can get regulatory approval. The fact is that so far all ideas about wholesale products with sliding or flexible billing scale have gone fubar in the very early stage of negotiating with Oftel/Ofcom and the blame for this is clearly with Oftel (now Ofcom).
To add to that, it is clearly the better company as far as customer service and reliability is concerned compared to any alternative in the UK. DSL outages in most areas are usually under 20-30 minutes per year which for example is clearly above the US average. Compare this to NTL or BullDog which happily has them in the days or even weeks range (speaking from experience). If you line is broken they fix it in usually less then 4h. Compare to NTLs 3-7 days. If you have a billing complaint they fix it within 20-30 minutes. Compare this to NTLs "NEVER". The only way to get misbilled money from them is though a lawsuit.
Yes, BT is bad, but a large chunk of its badness is due to the regulatory regime and the alternatives are much worse.
Slashdot Mirror
Fiasco of Carly?
Complete and utter bollockssssss.... Fiasco of the Don Capellas cretinomoron - yes. Carly - no. Whatever people say about her any Compaq/HP shop can clearly tell you that during her reign the quality of hardware shipped drastically improved. If you have bought HPaq hardware over the last 6+ years you can clearly see the day when the cost saving idiot Capellas reign started. The quality of build immediately dropped to rock bottom. Passive cooling was replaced by cheap mexican fans which failed in 3-5 months, spec of the average box in fact dropped instead of rising. 512MB memory limit on a desktop down from 784 and thermal throttle on the CPU which effectively made it slower then the older models.
After that, during Carly it recovered to a more or less sane state - custom designed motherboards matched to the case instead of cheap chinese OEM shit, passive cooling assisted by the main chassis fan (an old Compaq TM), etc.
Whatever you say while I am not a Carly fan, she actually fixed a lot of the shit created by Don Cappellas. It would have been nice to see her go to be replaced by someone to improve further. Instead, she is being replaced by a Don Cappelas clone which means that all of us who buy Compaq are likely to have to look for an alternative vendor once again.
If it is worth working. Every time I talk to an agent jobhunting for them he is trying to offer me a 20% salary cut. To add insult to injury they also have the nerve to ask if your current salary is "negotiable". I have started answering "Yes, if you would like to negotiate it in the right direction, in other words - UP".
No thanks. They may have a few really smart people like Delany on their staff, but with this rate of pay I somehow doubt that they are going to get anywhere near getting and retaining talent in general
Yes.
I know what I wrote. And I am a chemist by the way.
Water is a very complex substance. You have both van-der-vaals (weak) and polar (strong) interactions between the water molecules and dissolved substances which tend to be much more organised and complex then most average lamerz think. These can be altered by many factors including magnetic field and ionising radiation. After being altered they stay altered for days and sometimes weeks.
A.
No.
Realistically there are bottlenecks all over the place and out of them these 2 prevent nearly any computer from reaching 1G.
1. Interrupt handling bottleneck. Even with interrupt mitigation your typical pps value for a single CPU P4 is under 100 kps. It falls down to under 60 kps when using Intel dual CPUs (dunno about AMD or Via) or SMT due to the overly deep pipeline on the P4. That is way less then 1G for small packets.
2. IO bottleneck. Many motherboards have IO-to-memory speeds which are realistically way under 1G in total, usually around 600-700 Mbit.
No card can help for these 2 problems.
RTFA.
The reporter reports ground zero being exactly on the railway line between the Mitsubishi plant and a residential district.
The fact that it was censored in all military reports is another matter.
In btw, I suspect that this is the exact reason why McArthur had it squashed.
And its target being the Mitsubishi naval yard which was 5 miles off. To add to that it also landed smack on top of a residential district with more then 50% of the ground 0 being a civilian target.
It was a war crime. Same as Dresden and same as Hiroshima.
Seconded.
How the f*** could it be selective and proper if the two idiots who dropped it misread the landmarks and dropped it more then 2 miles off target. It was in fact intended to hit the naval yards 5 km downstream from where it was dropped.
Not, just there.
All Bulgarian, Cheh and most Caucasus (Russian and Georgian) SPAs are like this.
Radon containing water works miracles on arthritis, joints problems as well as many forms of eczema. While it usually fails to provide permanent cure it provides 3-4 months of relief or gives medications a better chance to work.
In btw, the feeling is weird... 20 minutes in a warm pool of such water makes you feel like your joints have started to melt. They feel like rubber.
The mechanism is still unclear, but it is not the Radon which is the active agent. It is the way its decay changes water properties.
If you are referring to the fact that Miller used to call his machines with different obsceneties in Russian like pizda (vagina), huy (dick), so on, I am well aware of that. None the less, sparc tree remains indicative of "quirky" hardware compared to ppc even after Millerisms have been cleaned out.
While both of them improve, Jo average speed of typing and speed of perception does not. As a result while the amount of data grows (flash, animations, video), the amount of items remains relatively constant (or grows at a much slowlier rate). Do not forget that the DOJ (or its equivalent elsewhere) can subpoena the data from the source or destination or both. Hence all it needs to see at the ISP level is that the data has been exchanged. Similarly, the fact that the data has been exchanged is sufficient to subpoena the content (Carnivore anyone?).
There is plenty of technology to do this now. No need for storage improvement. They can get it now and they are likely to get it.
Silliness???
...
As far as the kernel is concerned the number of profanities is a clear reflection of the quality of the underlying hardware. One of the things I do before buying new hardware is look at the comments in the linux kernel code. If they are like the ones you meet in the sun** architecture bit (it is the most profane part of the linux kernel) it may be a good idea to stay away.
For example just read the sunhme.c under drives/net. It is an absolute ROFL. Or arch/sparc/mm/ptrace.c
Yes. But 2h is roughly the time it takes to get alternative capacity running on a friend of mine basis. Been there, done that, hate fishermen.
You may be right as Sun has payed SCO to use their crap, but not to rerelease it under another license (and OSS to boot).
As anyone who has had to suffer from admin-ing OpenServer the idea of parallel non-init.d style startup launching is not new. In fact it is very old.
SCO had it first.
Before touching this "open" thing I will spend at least several days comparing this bit to the old OpenServer bit. And if they have as much in common as I suspect the best thing to do here will be to duck and wait for SCO legal missiles to land on a legitimate target.
In btw, that will be a nice way for McNealy to pay a bit more to SCO to top up his previous anti-linux contribution without causing undue screams in the community. All legit. Stolen code. Out of court settlement. Rather large one...
As a matter of fact, dear coward, my grandfather was a member of the russian group before he volunteered (it took him more then 5 tries to be reassigned) to follow his second profession and fight in the air and die as test pilot in the winter of 1942-43. If you are already that proficient in WW2 and Russian history you may have already guessed who (beware names are different). So, coming back to what is essentially a family history for me, the Russian effort was not espionage. At least not to a the extent traditionally assigned to it. Russian technology for graphite purification and reactor design were different. Bomb itself - yes, it was too similar to the American designs for everyone to become very suspicious. Getting to it and especially getting a working reactor - no.
There is something many WW2 historians have noticed.
Heisenberg was not an engineer and he did not get along with engineers. Typical of most European academentia at the time (and even now) . There were very few people to actually take the "ideas" and make them "tick" on the German nuclear team.
This is in sharp contrast with both American and Russian efforts which were done in a much more practical manner.
No.
But the devil has been rumoured to drive a snowplough to work. Apparently he is fixing bugs in autofs, mailman, hpa-tftp(d) [which in fact is a linux kernel UDP bug]. Add glibc compat module to this.
To name just a few (from my "what do I follow and have tried to fix locally" list).
Well, besides that the penetration of computers into more and more parts of our life is decreasing the qualification of an average IT professional. Many people do IT without understanding anything about what they do. As a result the vendor services model is bound not to decrease. It is bound to grow.
In btw, when I saw the first forbes article I bet that it was PR placement by LMV. The second article proves it.
FPS freak...
Real geeks have IJKL (and to lesser degree UOBN) worn out. AZ follow... But W, S, D - who cares about what weapon you are wielding, searching is best done via an artefact or intrinsic and D... WTF was D anyway...
Hmm... Circa 1997 80%+ of country MSFT ops ran their webservers on Linux or Solaris. The moment Netcraft published this and they became a laughing stock it was all migrated to IIS within 2 weeks. Similarly, I do not really believe in such posturing. MSFT is a marketing driven organisation and if their marketing decides that a specific instance of running Linux is bad for the current marketing campaigns there will be a big crater in its place in less then 5 seconds.
//begin sarcasm //end sarcasm
Or tactical planning simulator. How to prepare to blow your city council building in the most efficient manner. As a matter of fact it will be great if someone writes a map converter to load these into a decent flight simulator or into one of the first person shooters. And once again, quite usefull on planning your revenge on that pesky council clerk that did not give you planning permission for that nice cannon tower you were planning in your backyard
On a more serious note, I somhow have the suspicion that some data will be unavailable or deliberately distorted. I simply do not see MSN publishing 45 degree 1m resolution pictures of Vandenberg, Clarke Fields or anything similar.
Dependencies? In a debian upgrade? Linux is not R00tH4t and R00tH4t aint linux. I think you need to run apt-get install -y lib-get-a-clue (with all dependecies)
And this is the exact reason why any good SLQ based system must have means of integrity checking.
As someone who have been writing database stuff for 10+ years now, I get really pissed off when I see lunatics raving on Acid about ACID. ACID in itself is not enough.
You must have reference checking, offline integrity tests as well as ongoing online integrity test. Repeating your example a transaction for buying tickets for a holiday must insert a record in the Requests table, Tickets table, Holidays table, etc and you must have an offline tool (even better backround thread) which checks that all records are present. In fact for the same reason in a well designed system you must violate 3rd normal form and have the integrity checking tool use the redundant data as well. Another alternative is a state load and checksum across the state storing it back in at least two different places (once again breaking 3rd normal form).
If you do it this way you can get a working system even if ACID breaks (databases have bugs), you can recover if hardware breaks and most importantly you have a considerable level of fraud resistance.
Maxtor 30 and 40 low profile run very hot.
I have had so far 0 failure rate with them in servers with good cooling where they can be chilled down nicely to sub-30C due to the low profile which greatly improves airflow (knocking on wood here).
At the same time I have had 50%+ 1st year failure rate in Cubid mini ITX cases, 30%+ in Compaqs and other similar systems without dedicated disk fans. If their SMART sensors are to be believed they heat up to 42C+ in a HPaq and 50C+ in an ITX case.
If you run them in a desktop you must have all power management features enabled and must spindown the drive when not in use. If you do that, it will survive even in an mini ITX case. Otherwise your data will be toast. Literally.
Cisco has always been a champion in managing to ship a record number of bugs and still stay alive. Compare to them Microsoft is a stellar example of quality assurance.
Just look at the mixture of puke and shite they have for a TCP stack (do not mistake it with the forwarding engine which is a different thing). Every single TCP weakness reported in the last 3 years has cisco amidst the vulnerable list. Inadequate reaction to ICMPs, insecure resets as a reaction to bogus data, you name it.
Interestingly enough they are still out there despite all their bugs. My personal take on this is that they have managed to create create a stable ecosystem of certified "specialists" that do not know anything but cisco and do not buy anything but cisco. From there on it is not really relevant if it breaks.
Anyway, back on topic. Even a record number of bugs does not mean that a system is a usefull target for intrusion. Cisco is not such a target, because it is monolytic. You cannot easily transfer control from a bug to something that is usefull for you (some form of command line). There is nothing to "exec" so you have to figure out the address of an interpreter instance with enable access and manage to invoke it without buggering up the scheduler and the nearly insane memory allocation. Definitely non-trivial. Compare this to a unix platform where for a "good bug" you can execute a shell with under 30 bytes and the difference becomes quite obvious. Futher to that the CPU on an average cisco is outright flimsy so even after taking over one you cannot really use it to listen to network traffic. It has nearly no local storage so you cannot use it to stage an attack or as a zombie.
So on so fourth.
It is hard to hack and very useless. Once upon a time the answer to "I have root on this router, what can I do with it" was "You can shovel it up your arse". That is still the case.
In btw, I read the article and it does not compute. AFAIK Cisco (as a company) uses one-time passwords on all internet accessible systems. SecureID to be most exact. Used to be SKEY prior to 1998 (or 7?). So you can sniff passwords until you puke without getting anywhere. The article is missing a few steps on how the guy got from the bastion hosts to real stuff. Alternatively (and more likely) it is full of shit.
This means that you have not dealt with Telewest, Homechoice, Bulldog or any of the other "alternatives".
First, lets be clear. BT are in this for the money, not for the coolness. They have no intention on offering cool products dot-bomb style that do not bring profits.
Second, they may seem technologically backwater, but they are obliged by the UK regulatory regime to offer their products on a national basis. As a result if it takes to limit DSL to 512 (old Fujitsu linecards) or 1M (new linecards) to offer it for the coverage defined in the approved wholesale product they will do so without any regrets.
If you want BT to offer better products they happily will, provided that they can charge you for it and that the product can get regulatory approval. The fact is that so far all ideas about wholesale products with sliding or flexible billing scale have gone fubar in the very early stage of negotiating with Oftel/Ofcom and the blame for this is clearly with Oftel (now Ofcom).
To add to that, it is clearly the better company as far as customer service and reliability is concerned compared to any alternative in the UK. DSL outages in most areas are usually under 20-30 minutes per year which for example is clearly above the US average. Compare this to NTL or BullDog which happily has them in the days or even weeks range (speaking from experience). If you line is broken they fix it in usually less then 4h. Compare to NTLs 3-7 days. If you have a billing complaint they fix it within 20-30 minutes. Compare this to NTLs "NEVER". The only way to get misbilled money from them is though a lawsuit.
Yes, BT is bad, but a large chunk of its badness is due to the regulatory regime and the alternatives are much worse.