Less government regulation on mine safety and pollution? Probably going to be a problem once they start trying to deliver tons of rock from orbit, although possibly people trying to deliver tons of rock from orbit are in a strong negotiating position...
I was always under the "assumption" that the success of Unix had to do with the fact that it was written mainly in C so you could port it to any platform with a C compiler
UNIX emerged at a time when the cost of computer was far greater than the cost of the developer effort to write an OS. Writing an OS from scratch in assembly (or, occasionally, in Algol or similar) was a fairly common task at the time and the OS was often a differentiating feature (e.g. VM/370).
, that it supported a rich programming API even very early on,
Nope, it supported an incredibly limited API. No initial support for shared libraries, no support for structured files, no ACLs. It was very primitive in comparison to mainframe or minicomputer operating systems of the era.
that it was not bloated allowing it to not waste what was at the time extremely expensive computing resources,
One man's bloat is another man's useful feature set. Given that modern UNIX systems now copy almost all of that 'bloat', it seems likely that it was actually useful...
and that it actually supported multiple users.
Again, not a major selling point since all of its competitors did as well. It was only systems at the very low end like CP/M (which appeared a bit later) that didn't.
But everyone knows OS X isn't a real Unix. Lets face it, they do just enough to pay for their certification.
Everyone knows this? I didn't. Paying for their certification doesn't mean that they pay and get to use the UNIX brand, it means that they pay and get to publish the results of running the Single UNIX Specification test suite. The 'just enough' that they do means implementing the Single UNIX Specification well enough to pass all of the tests. For those of us who write low-level code, that means that OS X is one of the few platforms where code written to the spec is actually likely to work as expected, and in my experience it does a lot better than, for example, GNU/Linux (see locale support or things like lio_listio() for examples). Perhaps you could point to some things that a real UNIX does that OS X doesn't?
Optionally, they can put their own equipment in the telephone exchanges though this is generally limited to the larger of the (still pretty small) alternative ISPs
This is only true on exchanges with local loop unbundling. I'm not sure what percentage of exchanges now support LLU, but it's based on consumer demand (which is silly, because consumers don't demand it) and my mother's exchange only got it very recently.
The telephone line rental is totally separate from the broadband
And, worse, BT does not offer naked DSL, so unless you are on an LLU exchange and using an LLU supplier then you get to pay BT £14.60/month for a phone line that you might not use. If you don't use the landline then it's almost impossible for ADSL ISPs to compete with Virgin in areas with cable. Virgin will sell you a 30Mb/s connection for £22.50/month. Subtract the line rental from this and ISPs have to charge under £7.90/month for the Internet access to offer the same price. If you do use the phone line, they need to charge £13.80/month to be competitive. Given the BT OpenReach charges for backbone traffic, It's basically impossible for ISPs to offer these prices without very low bandwidth caps.
The upshot is that British Telecom was split into two companies
Not quite. BT OpenReach and BT Retail are both divisions of BT. The only constraint imposed by Ofcom is that BT OpenReach must provide access to third parties at the same price that it provides access to BT Retail. This does not, however, have to be at a rate that would allow BT Retail to be profitable. As long as BT Retail is just about breaking even, after the savings it makes sharing a lot of things like payroll with the rest of BT, it satisfies the regulators. And while OpenReach is raking in the cash, there is no incentive for BT to change this structure. Unfortunately, Virgin Media is upgrading its network, but not expanding it, so OpenReach retains a monopoly in all of the areas where Virgin hasn't got a presence.
Sure, and there are lots of even more clever ways of doing it. My point is that even without really trying - just with having two mail accounts, in a fairly generic configuration, in different countries you get a chain that gives the intelligence services nothing. The terrorists don't even need to be vaguely clever to do that, they just have to not use hotmail (actually, depending on intelligence sharing, using hotmail might also work - all that the UK side would know was that they both connected to the same US-based server farm as a million or so other people). If you post some nonsense in every Slashdot article with an easily recognisable heading (e.g. buy cheap viagra!) and then an encrypted message and someone else posts a reply using a similar scheme on an unrelated - but also high-traffic - messaging board, also in code, then no one will notice. On Slashdot, the post will be moderated down so most people won't even see it. People who do see it will just assume it's another spam bot.
40? Not really. 20 maybe, but in and before the '80s the term 'UNIX security' was something that users of things like VMS and OS/370 said while sniggering. A big part of the success of UNIX was that it ran on cheaper hardware than a real operating system...
do we want more privacy, accepting increased risks of criminal/terrorist acticity, or are we willing to trade off some privacy to get more security. This choice is political and should be democratically decided. Whether this applies to the internet or in other contexts actually
The problem is, we've not seen any real evidence that sacrificing privacy actually does result in increased security. Terrorists can easily use off-the-shelf tools like anonymous remailers, Tor and encryption and so the intelligence services don't get any information about who they're talking to or what they're saying even if they record and analyse 100% of UK Internet traffic. If a terrorist makes an encrypted SMTPS connection to a server in, say, China, that mail server makes (after a random delay) another encrypted connection to a mail server in, for example, Brazil, and then another terrorist collects the mail from the server in Brazil via IMAPS, then what can you learn? Very little unless you can monitor the entire Internet, and the Chinese probably don't want you to monitor their part any more than you want them to monitor yours.
Most counterterrorism operations get their intelligence from far more traditional sources.
The small ISPs did not become big ISPs, the incumbent telecoms and cable companies became big ISPs. There are still a lot of small ISPs, but they account for under 10% of the market between them. Virgin Media and BT control almost all of the physical infrastructure, and along with a handful of other big companies (e.g. Sky) also control the majority of the customer-facing side. Even if you go with AAISP, they're still using BT's network, so there's little they can do if BT starts snooping on the backbone.
Doesn't help if you've got a PowerPC Mac, of course, since it will only run 10.5, and is still fast enough for a lot of things, especially if it's something like a dual 2GHz G5 (although even a 1.5GHz G4 is faster than the old PC my mother still uses).
Still, I'm sure that these numbers will make someone at Oracle happy: at least half a million people still have machines set up to run Java applets...
Because Java might not be the best tool for the job? I know some Java developers find this a shocking idea, but it's really true: sometimes another language and environment is a better solution.
Those are the 1.6 docs. Java 1.7 also uses TimSort. Of course, since this was released in 2011, it's entirely possible that this means Oracle has been copying Android code...
Unlike the Google version, where they popularise an incompatible version, calling it Java and where it's possible that programs developed with Google's version would not operate correctly with Oracle (or anyone else's compliant) version?
I pay under the equivalent of $5 for my phone. It's prepay, most of my calls are incoming (so free) and I make outgoing calls when I'm in near WiFi over SIP (much cheaper for calling landlines).
Depends on the typesetting process. A lot of journals that are published by academics, rather than by publishing companies, generate the final version by just taking the camera-ready manuscripts from the authors and overlaying their own header and footer (with page numbers, paper names, and so on).
I'd agree on Diamond Age. Cryptonomicon was a bit dull. It rehashed a load of history and crypto theory that I knew already and squeezed a plot into the 50 or 60 pages left over. Read something like The Code Book instead if you want to learn about the history of cryptography - it's better written and more accurate, and doesn't try to shoehorn a weak plot into it.
I have a lot of science fiction from the '60s and the themes that run through pretty much all of it include:
Space travel will become cheap
Computers? What are they? Occasionally you get robots (basically human slaves, but made using technology) or mechanical brains. The few times you do get computers, they're massive things serving a whole continent or planet.
Nuclear war is inevitable.
The last point is the most important. Science fiction writers in the '60s had two preferred settings, a post apocalyptic world, or a world that has recovered from the aftermath of nuclear war and learned from the experience.
The documents look fantastic and the end-users don't need to know anything about LaTeX -- which really isn't something the average user can handle.
Let's face it - it's something that most users can't handle. It's a programming language with no concept of scoping and no structured flow control and no formal separation between content and markup or between semantic and presentation markup. I'd love to see a review of TeX by Dijkstra...
The only way to use LaTeX is via metaprogramming. I use a set of semantic markup tags in TeX syntax in chapter files and then a load of ugly goo in a document class and a preamble. When I want to generate ePub, I parse the semantic markup, ignore the preamble stuff, and just spit out corresponding semantic XHTML that I can then style with CSS.
Oh, I don't expect these organisations to act with integrity, I just thought that having someone with a reputation for being wrong 90% of the time and who talks such obvious nonsense on their side would be more of a liability. It would make as much sense as Obama paying for an endorsement form Kim Jong Il in the 2008 elections.
That depends on the terms of the agreement. In the case of such agreements with IBM, they usually agree not to bombard you with spurious patent lawsuits until you go bankrupt in exchange for being allowed to use all of your technology.
From the top of the screen to the top of the page, the default ribbon layout in Word uses THE SAME vertical space than the default menu+toolbars in open office writer.
But you can turn off all of the toolbars, and still have everything reachable form the menus in less mouse movement than from a hidden ribbon (although more from an unhidden ribbon).
2) There are keyboard shortcuts to every feature in the ribbon. Press Alt and follow the letters. This is more discoverable and provides more functionality.
Unlike the more conventional shortcuts, these are not side-effect free. They change the currently exposed tab on the ribbon. If you have your quick access tab open, and you save with alt-f-s (for example) instead of control-s, then you will now have the file menu open and need more mouse movement to return to the old state.
If I recall, IBM doesn't wield patents like..say.. Oracle
Let's look at a company that Oracle bought recently and their origins...
When Sun was starting up, they got a visit from the Nazgul[1] with seven IBM patents. They met the Sun team (only a handful of people then), who showed that there was enough prior art to invalidate some of the patents and that they weren't infringing on any of the others. Then Nazgul agreed that, no, indeed, Sun was not infringing on these seven patents and said 'would you like us to come back with some you are infringing?' Sun signed a cross-licensing deal with IBM.
You may recall that Microsoft became a bit more aggressive in chasing patent licensing in the last few years. This happened as a result of hiring an ex-IBM person to run their IP racketeering department.
In fact, if I had to think of one tech company best known for, shall we say, aggressive monetization of IP, then IBM would be my first pick.
[1] Wiki-annecdote: in spite of the fact that IBM's lawyers have been referred to as Nazgul since the late '70s, some random guy on Wikipedia has not heard this before, and so the Nazgul page no longer mentions this.
The other big problem with outsourcing to India is finding the competent Indians. It's much easier to find competent people if you can interview them in person than if they are a few hours away on a fast aeroplane. Unless you have a large corporate presence in India already, you're likely to get the people who couldn't get a job somewhere that does competent screening, even if you are paying above market rates. Add to that, the competent ones are not stupid: they quickly learn that they can consult one or two days a month charging close to western rates and live very comfortably...
You can actually do most of this in LaTeX too. My latest book is about Go and was written while the final spec was still in flux, so it got a lot of iterations through copyediting before the final version. I used latexdiff to produce PDFs indicating all of the changes since the previous version, including insertions and deletions. I have it integrated into my build system, so I can just specify a subversion revision number and have it give me a diff against that one. You can also use standard tools for revision control and for small changes just look at the raw diff without typesetting it.
OpenOffice does have change tracking, but somehow the performance is terrible. On a large document, with changes visible, on a 2.2GHz quad core i7, it took several seconds for each character I typed to show up.
Slashdot Mirror
Less government regulation on mine safety and pollution? Probably going to be a problem once they start trying to deliver tons of rock from orbit, although possibly people trying to deliver tons of rock from orbit are in a strong negotiating position...
I was always under the "assumption" that the success of Unix had to do with the fact that it was written mainly in C so you could port it to any platform with a C compiler
UNIX emerged at a time when the cost of computer was far greater than the cost of the developer effort to write an OS. Writing an OS from scratch in assembly (or, occasionally, in Algol or similar) was a fairly common task at the time and the OS was often a differentiating feature (e.g. VM/370).
, that it supported a rich programming API even very early on,
Nope, it supported an incredibly limited API. No initial support for shared libraries, no support for structured files, no ACLs. It was very primitive in comparison to mainframe or minicomputer operating systems of the era.
that it was not bloated allowing it to not waste what was at the time extremely expensive computing resources,
One man's bloat is another man's useful feature set. Given that modern UNIX systems now copy almost all of that 'bloat', it seems likely that it was actually useful...
and that it actually supported multiple users.
Again, not a major selling point since all of its competitors did as well. It was only systems at the very low end like CP/M (which appeared a bit later) that didn't.
But everyone knows OS X isn't a real Unix. Lets face it, they do just enough to pay for their certification.
Everyone knows this? I didn't. Paying for their certification doesn't mean that they pay and get to use the UNIX brand, it means that they pay and get to publish the results of running the Single UNIX Specification test suite. The 'just enough' that they do means implementing the Single UNIX Specification well enough to pass all of the tests. For those of us who write low-level code, that means that OS X is one of the few platforms where code written to the spec is actually likely to work as expected, and in my experience it does a lot better than, for example, GNU/Linux (see locale support or things like lio_listio() for examples). Perhaps you could point to some things that a real UNIX does that OS X doesn't?
Optionally, they can put their own equipment in the telephone exchanges though this is generally limited to the larger of the (still pretty small) alternative ISPs
This is only true on exchanges with local loop unbundling. I'm not sure what percentage of exchanges now support LLU, but it's based on consumer demand (which is silly, because consumers don't demand it) and my mother's exchange only got it very recently.
The telephone line rental is totally separate from the broadband
And, worse, BT does not offer naked DSL, so unless you are on an LLU exchange and using an LLU supplier then you get to pay BT £14.60/month for a phone line that you might not use. If you don't use the landline then it's almost impossible for ADSL ISPs to compete with Virgin in areas with cable. Virgin will sell you a 30Mb/s connection for £22.50/month. Subtract the line rental from this and ISPs have to charge under £7.90/month for the Internet access to offer the same price. If you do use the phone line, they need to charge £13.80/month to be competitive. Given the BT OpenReach charges for backbone traffic, It's basically impossible for ISPs to offer these prices without very low bandwidth caps.
The upshot is that British Telecom was split into two companies
Not quite. BT OpenReach and BT Retail are both divisions of BT. The only constraint imposed by Ofcom is that BT OpenReach must provide access to third parties at the same price that it provides access to BT Retail. This does not, however, have to be at a rate that would allow BT Retail to be profitable. As long as BT Retail is just about breaking even, after the savings it makes sharing a lot of things like payroll with the rest of BT, it satisfies the regulators. And while OpenReach is raking in the cash, there is no incentive for BT to change this structure. Unfortunately, Virgin Media is upgrading its network, but not expanding it, so OpenReach retains a monopoly in all of the areas where Virgin hasn't got a presence.
Sure, and there are lots of even more clever ways of doing it. My point is that even without really trying - just with having two mail accounts, in a fairly generic configuration, in different countries you get a chain that gives the intelligence services nothing. The terrorists don't even need to be vaguely clever to do that, they just have to not use hotmail (actually, depending on intelligence sharing, using hotmail might also work - all that the UK side would know was that they both connected to the same US-based server farm as a million or so other people). If you post some nonsense in every Slashdot article with an easily recognisable heading (e.g. buy cheap viagra!) and then an encrypted message and someone else posts a reply using a similar scheme on an unrelated - but also high-traffic - messaging board, also in code, then no one will notice. On Slashdot, the post will be moderated down so most people won't even see it. People who do see it will just assume it's another spam bot.
40? Not really. 20 maybe, but in and before the '80s the term 'UNIX security' was something that users of things like VMS and OS/370 said while sniggering. A big part of the success of UNIX was that it ran on cheaper hardware than a real operating system...
do we want more privacy, accepting increased risks of criminal/terrorist acticity, or are we willing to trade off some privacy to get more security. This choice is political and should be democratically decided. Whether this applies to the internet or in other contexts actually
The problem is, we've not seen any real evidence that sacrificing privacy actually does result in increased security. Terrorists can easily use off-the-shelf tools like anonymous remailers, Tor and encryption and so the intelligence services don't get any information about who they're talking to or what they're saying even if they record and analyse 100% of UK Internet traffic. If a terrorist makes an encrypted SMTPS connection to a server in, say, China, that mail server makes (after a random delay) another encrypted connection to a mail server in, for example, Brazil, and then another terrorist collects the mail from the server in Brazil via IMAPS, then what can you learn? Very little unless you can monitor the entire Internet, and the Chinese probably don't want you to monitor their part any more than you want them to monitor yours.
Most counterterrorism operations get their intelligence from far more traditional sources.
The small ISPs did not become big ISPs, the incumbent telecoms and cable companies became big ISPs. There are still a lot of small ISPs, but they account for under 10% of the market between them. Virgin Media and BT control almost all of the physical infrastructure, and along with a handful of other big companies (e.g. Sky) also control the majority of the customer-facing side. Even if you go with AAISP, they're still using BT's network, so there's little they can do if BT starts snooping on the backbone.
Doesn't help if you've got a PowerPC Mac, of course, since it will only run 10.5, and is still fast enough for a lot of things, especially if it's something like a dual 2GHz G5 (although even a 1.5GHz G4 is faster than the old PC my mother still uses).
Still, I'm sure that these numbers will make someone at Oracle happy: at least half a million people still have machines set up to run Java applets...
Because Java might not be the best tool for the job? I know some Java developers find this a shocking idea, but it's really true: sometimes another language and environment is a better solution.
Those are the 1.6 docs. Java 1.7 also uses TimSort. Of course, since this was released in 2011, it's entirely possible that this means Oracle has been copying Android code...
Unlike the Google version, where they popularise an incompatible version, calling it Java and where it's possible that programs developed with Google's version would not operate correctly with Oracle (or anyone else's compliant) version?
I pay under the equivalent of $5 for my phone. It's prepay, most of my calls are incoming (so free) and I make outgoing calls when I'm in near WiFi over SIP (much cheaper for calling landlines).
Depends on the typesetting process. A lot of journals that are published by academics, rather than by publishing companies, generate the final version by just taking the camera-ready manuscripts from the authors and overlaying their own header and footer (with page numbers, paper names, and so on).
I'd agree on Diamond Age. Cryptonomicon was a bit dull. It rehashed a load of history and crypto theory that I knew already and squeezed a plot into the 50 or 60 pages left over. Read something like The Code Book instead if you want to learn about the history of cryptography - it's better written and more accurate, and doesn't try to shoehorn a weak plot into it.
Some people can't handle epics.
I agree. Neal Stephenson seems to be one of them.
The last point is the most important. Science fiction writers in the '60s had two preferred settings, a post apocalyptic world, or a world that has recovered from the aftermath of nuclear war and learned from the experience.
The documents look fantastic and the end-users don't need to know anything about LaTeX -- which really isn't something the average user can handle.
Let's face it - it's something that most users can't handle. It's a programming language with no concept of scoping and no structured flow control and no formal separation between content and markup or between semantic and presentation markup. I'd love to see a review of TeX by Dijkstra...
The only way to use LaTeX is via metaprogramming. I use a set of semantic markup tags in TeX syntax in chapter files and then a load of ugly goo in a document class and a preamble. When I want to generate ePub, I parse the semantic markup, ignore the preamble stuff, and just spit out corresponding semantic XHTML that I can then style with CSS.
Oh, I don't expect these organisations to act with integrity, I just thought that having someone with a reputation for being wrong 90% of the time and who talks such obvious nonsense on their side would be more of a liability. It would make as much sense as Obama paying for an endorsement form Kim Jong Il in the 2008 elections.
That depends on the terms of the agreement. In the case of such agreements with IBM, they usually agree not to bombard you with spurious patent lawsuits until you go bankrupt in exchange for being allowed to use all of your technology.
Apple has, for the past four years, been trying to do what IBM has done successfully for the last forty.
From the top of the screen to the top of the page, the default ribbon layout in Word uses THE SAME vertical space than the default menu+toolbars in open office writer.
But you can turn off all of the toolbars, and still have everything reachable form the menus in less mouse movement than from a hidden ribbon (although more from an unhidden ribbon).
2) There are keyboard shortcuts to every feature in the ribbon. Press Alt and follow the letters. This is more discoverable and provides more functionality.
Unlike the more conventional shortcuts, these are not side-effect free. They change the currently exposed tab on the ribbon. If you have your quick access tab open, and you save with alt-f-s (for example) instead of control-s, then you will now have the file menu open and need more mouse movement to return to the old state.
If I recall, IBM doesn't wield patents like ..say .. Oracle
Let's look at a company that Oracle bought recently and their origins...
When Sun was starting up, they got a visit from the Nazgul[1] with seven IBM patents. They met the Sun team (only a handful of people then), who showed that there was enough prior art to invalidate some of the patents and that they weren't infringing on any of the others. Then Nazgul agreed that, no, indeed, Sun was not infringing on these seven patents and said 'would you like us to come back with some you are infringing?' Sun signed a cross-licensing deal with IBM.
You may recall that Microsoft became a bit more aggressive in chasing patent licensing in the last few years. This happened as a result of hiring an ex-IBM person to run their IP racketeering department.
In fact, if I had to think of one tech company best known for, shall we say, aggressive monetization of IP, then IBM would be my first pick.
[1] Wiki-annecdote: in spite of the fact that IBM's lawyers have been referred to as Nazgul since the late '70s, some random guy on Wikipedia has not heard this before, and so the Nazgul page no longer mentions this.
The other big problem with outsourcing to India is finding the competent Indians. It's much easier to find competent people if you can interview them in person than if they are a few hours away on a fast aeroplane. Unless you have a large corporate presence in India already, you're likely to get the people who couldn't get a job somewhere that does competent screening, even if you are paying above market rates. Add to that, the competent ones are not stupid: they quickly learn that they can consult one or two days a month charging close to western rates and live very comfortably...
You can be arrested for distributing child pornography if you are in the USA. See Slashdot back stories for a specific example...
You can actually do most of this in LaTeX too. My latest book is about Go and was written while the final spec was still in flux, so it got a lot of iterations through copyediting before the final version. I used latexdiff to produce PDFs indicating all of the changes since the previous version, including insertions and deletions. I have it integrated into my build system, so I can just specify a subversion revision number and have it give me a diff against that one. You can also use standard tools for revision control and for small changes just look at the raw diff without typesetting it.
OpenOffice does have change tracking, but somehow the performance is terrible. On a large document, with changes visible, on a 2.2GHz quad core i7, it took several seconds for each character I typed to show up.