Computers cannot lie because they do not deal in truth.
They most certainly do, for many reasonable definitions of "truth". In particular, any definition which does not resort to an appeal to some ineffable essence should be capturable by a computable formal system.
Of course, that will be limited by the limitations of the formal system in question, and any such system is either limited in power (compared to some other describable systems) or limited by incompleteness. Doxastic logic is Gödel-complete, as is any system isomorphic to it, obviously.
Your general point re intention is well taken. However:
A lie is an intentional deception. If you have a computer that can lie to you, it must have passed the Turing test a long time ago and may even be self-aware by now.
The Turing Test has nothing to do with intention, sentience, or sapience.
It's a philosophical proposal to illustrate the thesis that the attribute "thinking" should be applied to an entity based on pragmatic analysis - what programmers might call "black box" testing. It can be contrasted with, for example, Searle's Chinese Room, which is a philosophical proposal to argue an opposing position: that the attribute "thinking" implies some essence beyond what's observable from outside the system.
As various people (such as Robert French) have pointed out, the Turing Test is not in fact a useful test for thinking - and there's little reason to believe Turing himself thought it would be. It's purely a mental exercise: if you accept the thesis of the TT, then you accept the idea of mechanical thought; if you don't accept it, then what's your counterargument?
Of course various people - Searle, as noted above, and Penrose, and various others - have presented counterarguments. Some of them (e.g. Searle) accept the possibility of mechanical thought, but not the TT; others don't believe in mechanical thought, or (e.g. Penrose) believe that it requires a system strictly more powerful than a finite-space, finite-time approximation of a Turing Machine.
However, given all that, we might posit, say, a machine which is capable of thought and intention, but can only communicate on a single topic. Such a system would fail the TT against any reasonable set of judges, but would still be capable of lying.
Indeed. The "bathtub full of ice" urban legend might be good for some pre-teen campfire stories, but it makes no sense. Why would an organlegger leave the victim alive? There's no advantage in doing so, and considerable cost.
But perhaps jellomizer can cite a reputable source backing his claim. (N.B. Lame Miguel Ferrer movies are not such a source.)
after all we already stole "code" from them to now mean software.
As usual we are in the dim twilight of Poe's Law here, but just in case you were serious: no, no we did not. "Code" does not exclusively mean "secret code", and is not solely the domain of cryptology.
Etymologically, "code" means simply "writing" or "book". As a term of art, its use for systems of formal expression, and documents using such systems, is at least as old as its cryptological use.
The black & white set we kids used had those dials, but if memory serves, the fancy color set my parents had in the bedroom had dispensed with the fine-tuning knobs.
I watched a fair bit of programming in black & white, and I dare say I enjoyed it about as much as I enjoy anything on the HD set these days. By pretty much any measure, there's more quality programming available today - because there's more programming in total - but there was some good stuff then, too.
And for me it was really more about the social experience of watching with other people. I don't think we need Yet Another rant about smartphones and the death of shared experience, though.
I recall using multiple virtual desktops under X-Windows back in the 1980s
Hmm. I'm trying to think of an X11 window manager with virtual desktop support from the '80s. vtwm was released in 1990, so it doesn't quite qualify. Same is true of tvtwm, I believe (and I think that also was a single virtual desktop like swm, as described below).
uwm (the original, non-reparenting, beloved Ultrix window manager), mwm (Motif), awm (Ardent), cwm (IBM's Cambridge window manager for Project Athena)... none of those offered multiple virtual desktops, as far as I can remember.
swm (Solbourne) had a single virtual desktop, larger than the screen; the screen was a viewport onto it, and you could pan the viewport around. Not multiple desktops, though.
Pretty sure the window manager I wrote circa '99 (now long lost) didn't have virtual desktops either. It wasn't really a big thing then. Many people running X11 had multihead workstations.
Some OSes had virtual console devices and let you start multiple copies of X11 on them and hot-key between them. Obviously that's not the same as a virtualizing window manager, though. And I can't offhand think of any in the '80s that did this, though maybe IRIX did? It seems like the sort of thing SGI would implement for the lulz. (Like fsn. "This is UNIX! I know this!")
Ah, X11 in the '80s and early '90s. So many options, and so much more interesting than Gnome-versus-KDE.
All extant real Brainfuck implementations have finite-size cells, so unless the implementation specifically traps wrapping, you can dispense with either the increment or decrement operator. Similarly, many have fixed-length tapes, which let you omit either move-right or move-left.
Unlambda is another tarpit language with a bunch of syntactic sugar for the lazy programmer. (Who needs i when you have ``skk? No one, that's who.)
Even ETA programs could really get by without E.
I'm sure there's at least one tarpit language that needs every feature to be Turing-complete, though.
House of Cards will begin in 22 minutes after some buffering
Back in the Dark Ages we watched our chosen television programs when they were broadcast, and not at our convenience. And while many died horribly as a result, a surprising number of us lived to tell the tale.
There may well be problems with ISPs prioritizing traffic by origin or type - and I'm certainly no fan of Pai - but this particular one doesn't bother me a bit. While I enjoy the occasional streamed content, I wouldn't miss it if it were gone.
wonder how much it costs to buy a "story" in wired.
If it costs you very much, you're doing it wrong. Ryan Holiday lays out the process nicely in Trust Me, I'm Lying. Pretty much anyone should be able to get pretty much anything onto a major media site with a little work and little or no money.
There probably never was a Golden Age of journalism (remember the Maine?), but this sure ain't it. The economics do not favor substantial, researched reporting.
Hey, Pierce: you are not everyone. Not everyone is you. Don't tell us what "we all" do.
What a jackass.
Don't get me wrong: I'm happy he likes FQ. I'm happy it looks like it might be serious competition for Chrome (which I've never liked, even without accounting for Google's addiction to spying on users). But, Christ, is it really so hard for the current generation of journalists to distinguish their subjective experience from objective fact, and learn that people are different?
"Satoshi-level crypto knowledge" isn't a very high barrier. Bitcoin was assembled from well-known algorithms and protocols. See Narayanan and Clark, "Bitcoin's Academic Pedigree".
The innovation by the inventor(s) of Bitcoin ("Satoshi") was to put those pieces together in a particular way, under a catchy name, at the right time.
Which is what most inventors do, after all; I don't want to diminish the achievement. It's not every day that someone cobbles together something a bit different out of well-known ideas and it becomes a big industry phenomenon.
But it's quite a few days.
(So could it have been Musk? It's not impossible, but it doesn't strike me as very likely.)
Hell, we're even seeing nerd slactivism here. Didn't you read the summary? There's a change.org petition! Who knows what's next? Trenchant blog posts, snarky listicles, boycotting RT for a few days... the (minimal-effort) sky's the limit.
My sense is that Justice League will forever be remembered as the cultural watershed moment that made annoyed fans a force to be reckoned with.
Lighting control, security, and whole house audio are areas where it clearly is valuable
"Clearly"? That all sounds dreadful to me. I'd pay extra not to have those "features".
Fortunately I don't have to, but it may yet become difficult to avoid all this crap. Our television died this week, and when I went to replace it, I found exactly one model for sale that didn't have idiotic, pointless, vulnerability-ridden "smart" features. (There were probably others that would work as conventional sets without network access, but of course it's impossible to tell from the box whether they'll refuse to work if not allowed on the net.)
What's wrong with you? Have you considered that perhaps some people have priorities that are different from yours?
a couple of UI quirks
I don't care about quirks. It has a new UI. That's a bad start right there.
this is _the best_ UI that FF has had in *years*
I'm sure they're glad you think so. Some of us do not. Yes, Australis is ghastly, but that's why we have Classic Theme Restorer. With FFQ, we don't even have that.
Tabs are threaded, something folks around here have been pleading for for _years_.
Folks here, and elsewhere, have also been pleading to have their extensions supported. To have the freedom to continuing customizing the UI. To be spared a bunch of inane UI changes just to get security fixes.
it is true that this browser is fast
I don't care. I frequently have four or more windows with a hundred or more tabs open, and my shit Dell work laptop's CPU is currently throttled by Dell's idiotic, broken power-supply DRM. But pre-Quantum FF remains fast enough. And I'll go out on a limb and hypothesize that there are other people who do not give a shit that FFQ is faster.
If you are a geek, you should be rooting for Firefox
I used to. I've used Firefox as long as it's existed, on multiple computers and platforms. I used Netscape Communicator before that, and Mosaic before that. Over the years I've tried other browsers, either as my default or (more commonly) to test something with them. I didn't care for any of them.
But Mozilla seems determined to alienate long-time FF users, and frankly this is the final blow for me. I'm switching to Pale Moon. And you know what? That won't force me to use either the advertiser or the monopolist.
If PM dies, I'll just grab the final stable branch and maintain the fucker myself. I've poked at the FF source before. People make a big deal over how complicated it is, but I've dealt with worse.
Cx, (or Go or Rust, or any other fad damned thing) will not replace C.
Certainly not any time soon. History suggests that once a programming language reaches a certain point of saturation, its lifetime is going to be very long. FORTRAN (now Fortran), LISP, and COBOL both still have large active bases and new development.
There have been other attempts at "better C", such as D. I don't hold out any hope for this Cx. Sure, go ahead and create another language; it's a useful mental exercise if nothing else. But it's not going to displace C.
Every one of his victims thought they could get something for nothing. They thought they could get unrealistic returns.
My impression is that many of his victims had no idea what they were doing; they were just people with money to burn who were told by their acquaintances to "invest with this guy". Sure, a lot - perhaps most - of Madoff's investors thought they would make unrealistic returns (and no doubt some did; that's what happens with a Ponzi scheme), but I think that in many cases it was simply a matter of following the crowd. People jump on fads.
The same was certainly true of the original Ponzi scheme. Some historians think even Ponzi himself didn't realize it was a scam - his bookkeeping was apparently terrible to the point of nonexistent, and he just had piles of cash that he'd dip into when an investor wanted to cash out. (Which makes it a rather odd sort of scam, though since Ponzi quickly attracted "advisors" who most certainly did know what was going on, it could hardly be considered an accident.)
Arial is a cheap knock off of Helvetica, it was chosen specifically because if you're not that interested in fonts it looks almost identical.
Not exactly. First, while Arial is a cheap knock-off of Helvetica, in the sense of being developed specifically to be a cheaper alternative, it was largely based on Monotype Grotesque, which predates Helvetica by three decades. Helvetica, despite its inflated reputation as some kind of wonder-face, is a pretty basic humanist grotesque. It was hardly groundbreaking; it was just a bit nicer than its competitors.
And Arial wasn't designed to be mistaken for Helvetica, but to be substituted for it. The letterforms have the same metrics so that a document intended to be printed in Helvetica can be printed in Arial without changing the layout.
The point, of course, was to avoid paying for a Helvetica license. Kind of like, oh, OpenOffice. Or OpenJDK. Or aftermarket car parts. Or any of a zillion other cheaper-substitute technologies.
Perhaps, if the "most widely deployed" claim were in any way even close to being true.
But it isn't. It's complete crap. MINIX's footprint in embedded controllers is still probably less than Linux's, and its overall footprint is dwarfed by Android.
I'm willing to wager there are more mobile phones in the world than intel ME enabled PCs at this point.
A quick, and entirely unverified, web search suggests Android devices (mostly not running on Intel ME-equipped CPUs) sell at a rate of around 1.4B per year, while Intel CPUs sell around 400M per year. So, yeah, MINIX is almost certainly way behind Android.
I'm not sure who first came up with that "most widely deployed" claim - I've seen it in a number of articles about ME and MINIX - but I doubt it's based on anything more than someone's wild-ass supposition.
Novell isn't really a company anymore, so much as a brand under Micro Focus (who bought Attachmate, who bought Novell).
Yes. As far as I know, that means we (Micro Focus) still own the copyright on the UNIX sources that Novell got from USL. In 2010, Attachmate confirmed that it intended to keep them after the merger; and while I don't remember seeing anything official, there are various stories online that refer to MF as "the current owner of UNIX".
Novell transferred the UNIX trademark and the rights to the UNIX specification to The Open Group, for use in the Single UNIX Specification.
Not very efficient. Simply announce a line of new Green Needles, a sustainable product made from dried herbaceous plant matter. Now you've reduced the problem to finding a needle in a needle stack.
Security is always last when implementing a new piece of software.
Not always. There are organizations that implement an SDLC correctly, with secure-development practices applied throughout the development process. And guess what? That leads to much lower security defect rates in new code, and gradually whittles away at security technical debt in old code.
Of course, the fact that your post (which, let's face it, is a cliche) was voted up to 5 Insightful shows that most places are still not implementing an SDLC, or they're doing it wrong.
Slashdot Mirror
Computers cannot lie because they do not deal in truth.
They most certainly do, for many reasonable definitions of "truth". In particular, any definition which does not resort to an appeal to some ineffable essence should be capturable by a computable formal system.
Of course, that will be limited by the limitations of the formal system in question, and any such system is either limited in power (compared to some other describable systems) or limited by incompleteness. Doxastic logic is Gödel-complete, as is any system isomorphic to it, obviously.
Don't they teach this stuff in school anymore?
Your general point re intention is well taken. However:
A lie is an intentional deception. If you have a computer that can lie to you, it must have passed the Turing test a long time ago and may even be self-aware by now.
The Turing Test has nothing to do with intention, sentience, or sapience.
It's a philosophical proposal to illustrate the thesis that the attribute "thinking" should be applied to an entity based on pragmatic analysis - what programmers might call "black box" testing. It can be contrasted with, for example, Searle's Chinese Room, which is a philosophical proposal to argue an opposing position: that the attribute "thinking" implies some essence beyond what's observable from outside the system.
As various people (such as Robert French) have pointed out, the Turing Test is not in fact a useful test for thinking - and there's little reason to believe Turing himself thought it would be. It's purely a mental exercise: if you accept the thesis of the TT, then you accept the idea of mechanical thought; if you don't accept it, then what's your counterargument?
Of course various people - Searle, as noted above, and Penrose, and various others - have presented counterarguments. Some of them (e.g. Searle) accept the possibility of mechanical thought, but not the TT; others don't believe in mechanical thought, or (e.g. Penrose) believe that it requires a system strictly more powerful than a finite-space, finite-time approximation of a Turing Machine.
However, given all that, we might posit, say, a machine which is capable of thought and intention, but can only communicate on a single topic. Such a system would fail the TT against any reasonable set of judges, but would still be capable of lying.
Indeed. The "bathtub full of ice" urban legend might be good for some pre-teen campfire stories, but it makes no sense. Why would an organlegger leave the victim alive? There's no advantage in doing so, and considerable cost.
But perhaps jellomizer can cite a reputable source backing his claim. (N.B. Lame Miguel Ferrer movies are not such a source.)
after all we already stole "code" from them to now mean software.
As usual we are in the dim twilight of Poe's Law here, but just in case you were serious: no, no we did not. "Code" does not exclusively mean "secret code", and is not solely the domain of cryptology.
Etymologically, "code" means simply "writing" or "book". As a term of art, its use for systems of formal expression, and documents using such systems, is at least as old as its cryptological use.
The black & white set we kids used had those dials, but if memory serves, the fancy color set my parents had in the bedroom had dispensed with the fine-tuning knobs.
I watched a fair bit of programming in black & white, and I dare say I enjoyed it about as much as I enjoy anything on the HD set these days. By pretty much any measure, there's more quality programming available today - because there's more programming in total - but there was some good stuff then, too.
And for me it was really more about the social experience of watching with other people. I don't think we need Yet Another rant about smartphones and the death of shared experience, though.
I recall using multiple virtual desktops under X-Windows back in the 1980s
Hmm. I'm trying to think of an X11 window manager with virtual desktop support from the '80s. vtwm was released in 1990, so it doesn't quite qualify. Same is true of tvtwm, I believe (and I think that also was a single virtual desktop like swm, as described below).
uwm (the original, non-reparenting, beloved Ultrix window manager), mwm (Motif), awm (Ardent), cwm (IBM's Cambridge window manager for Project Athena) ... none of those offered multiple virtual desktops, as far as I can remember.
swm (Solbourne) had a single virtual desktop, larger than the screen; the screen was a viewport onto it, and you could pan the viewport around. Not multiple desktops, though.
Pretty sure the window manager I wrote circa '99 (now long lost) didn't have virtual desktops either. It wasn't really a big thing then. Many people running X11 had multihead workstations.
Some OSes had virtual console devices and let you start multiple copies of X11 on them and hot-key between them. Obviously that's not the same as a virtualizing window manager, though. And I can't offhand think of any in the '80s that did this, though maybe IRIX did? It seems like the sort of thing SGI would implement for the lulz. (Like fsn. "This is UNIX! I know this!")
Ah, X11 in the '80s and early '90s. So many options, and so much more interesting than Gnome-versus-KDE.
All extant real Brainfuck implementations have finite-size cells, so unless the implementation specifically traps wrapping, you can dispense with either the increment or decrement operator. Similarly, many have fixed-length tapes, which let you omit either move-right or move-left.
Unlambda is another tarpit language with a bunch of syntactic sugar for the lazy programmer. (Who needs i when you have ``skk? No one, that's who.)
Even ETA programs could really get by without E.
I'm sure there's at least one tarpit language that needs every feature to be Turing-complete, though.
House of Cards will begin in 22 minutes after some buffering
Back in the Dark Ages we watched our chosen television programs when they were broadcast, and not at our convenience. And while many died horribly as a result, a surprising number of us lived to tell the tale.
There may well be problems with ISPs prioritizing traffic by origin or type - and I'm certainly no fan of Pai - but this particular one doesn't bother me a bit. While I enjoy the occasional streamed content, I wouldn't miss it if it were gone.
wonder how much it costs to buy a "story" in wired.
If it costs you very much, you're doing it wrong. Ryan Holiday lays out the process nicely in Trust Me, I'm Lying. Pretty much anyone should be able to get pretty much anything onto a major media site with a little work and little or no money.
There probably never was a Golden Age of journalism (remember the Maine?), but this sure ain't it. The economics do not favor substantial, researched reporting.
or the kids will quickly learn to bypass the blocks
Well, that would be constructive, anyway. So parental controls aren't completely worthless.
the multi-device, ultra-mobile lives we all lead
Hey, Pierce: you are not everyone. Not everyone is you. Don't tell us what "we all" do.
What a jackass.
Don't get me wrong: I'm happy he likes FQ. I'm happy it looks like it might be serious competition for Chrome (which I've never liked, even without accounting for Google's addiction to spying on users). But, Christ, is it really so hard for the current generation of journalists to distinguish their subjective experience from objective fact, and learn that people are different?
"Satoshi-level crypto knowledge" isn't a very high barrier. Bitcoin was assembled from well-known algorithms and protocols. See Narayanan and Clark, "Bitcoin's Academic Pedigree".
The innovation by the inventor(s) of Bitcoin ("Satoshi") was to put those pieces together in a particular way, under a catchy name, at the right time.
Which is what most inventors do, after all; I don't want to diminish the achievement. It's not every day that someone cobbles together something a bit different out of well-known ideas and it becomes a big industry phenomenon.
But it's quite a few days.
(So could it have been Musk? It's not impossible, but it doesn't strike me as very likely.)
But - but - nerds! Raging! This is unprecedented.
Hell, we're even seeing nerd slactivism here. Didn't you read the summary? There's a change.org petition! Who knows what's next? Trenchant blog posts, snarky listicles, boycotting RT for a few days ... the (minimal-effort) sky's the limit.
My sense is that Justice League will forever be remembered as the cultural watershed moment that made annoyed fans a force to be reckoned with.
Lighting control, security, and whole house audio are areas where it clearly is valuable
"Clearly"? That all sounds dreadful to me. I'd pay extra not to have those "features".
Fortunately I don't have to, but it may yet become difficult to avoid all this crap. Our television died this week, and when I went to replace it, I found exactly one model for sale that didn't have idiotic, pointless, vulnerability-ridden "smart" features. (There were probably others that would work as conventional sets without network access, but of course it's impossible to tell from the box whether they'll refuse to work if not allowed on the net.)
Jesus, what is wrong with you people?
What's wrong with you? Have you considered that perhaps some people have priorities that are different from yours?
a couple of UI quirks
I don't care about quirks. It has a new UI. That's a bad start right there.
this is _the best_ UI that FF has had in *years*
I'm sure they're glad you think so. Some of us do not. Yes, Australis is ghastly, but that's why we have Classic Theme Restorer. With FFQ, we don't even have that.
Tabs are threaded, something folks around here have been pleading for for _years_.
Folks here, and elsewhere, have also been pleading to have their extensions supported. To have the freedom to continuing customizing the UI. To be spared a bunch of inane UI changes just to get security fixes.
it is true that this browser is fast
I don't care. I frequently have four or more windows with a hundred or more tabs open, and my shit Dell work laptop's CPU is currently throttled by Dell's idiotic, broken power-supply DRM. But pre-Quantum FF remains fast enough. And I'll go out on a limb and hypothesize that there are other people who do not give a shit that FFQ is faster.
If you are a geek, you should be rooting for Firefox
I used to. I've used Firefox as long as it's existed, on multiple computers and platforms. I used Netscape Communicator before that, and Mosaic before that. Over the years I've tried other browsers, either as my default or (more commonly) to test something with them. I didn't care for any of them.
But Mozilla seems determined to alienate long-time FF users, and frankly this is the final blow for me. I'm switching to Pale Moon. And you know what? That won't force me to use either the advertiser or the monopolist.
If PM dies, I'll just grab the final stable branch and maintain the fucker myself. I've poked at the FF source before. People make a big deal over how complicated it is, but I've dealt with worse.
Cx, (or Go or Rust, or any other fad damned thing) will not replace C.
Certainly not any time soon. History suggests that once a programming language reaches a certain point of saturation, its lifetime is going to be very long. FORTRAN (now Fortran), LISP, and COBOL both still have large active bases and new development.
There have been other attempts at "better C", such as D. I don't hold out any hope for this Cx. Sure, go ahead and create another language; it's a useful mental exercise if nothing else. But it's not going to displace C.
I will say that it is tedious to constantly be passing around the max size parameter to every little function
It certainly is! And that's why C supports compound data types.
because that's the only way for a C function to safely deal with pointer data
True, if you have no idea how to use the language properly.
Every one of his victims thought they could get something for nothing. They thought they could get unrealistic returns.
My impression is that many of his victims had no idea what they were doing; they were just people with money to burn who were told by their acquaintances to "invest with this guy". Sure, a lot - perhaps most - of Madoff's investors thought they would make unrealistic returns (and no doubt some did; that's what happens with a Ponzi scheme), but I think that in many cases it was simply a matter of following the crowd. People jump on fads.
The same was certainly true of the original Ponzi scheme. Some historians think even Ponzi himself didn't realize it was a scam - his bookkeeping was apparently terrible to the point of nonexistent, and he just had piles of cash that he'd dip into when an investor wanted to cash out. (Which makes it a rather odd sort of scam, though since Ponzi quickly attracted "advisors" who most certainly did know what was going on, it could hardly be considered an accident.)
Arial is a cheap knock off of Helvetica, it was chosen specifically because if you're not that interested in fonts it looks almost identical.
Not exactly. First, while Arial is a cheap knock-off of Helvetica, in the sense of being developed specifically to be a cheaper alternative, it was largely based on Monotype Grotesque, which predates Helvetica by three decades. Helvetica, despite its inflated reputation as some kind of wonder-face, is a pretty basic humanist grotesque. It was hardly groundbreaking; it was just a bit nicer than its competitors.
And Arial wasn't designed to be mistaken for Helvetica, but to be substituted for it. The letterforms have the same metrics so that a document intended to be printed in Helvetica can be printed in Arial without changing the layout.
The point, of course, was to avoid paying for a Helvetica license. Kind of like, oh, OpenOffice. Or OpenJDK. Or aftermarket car parts. Or any of a zillion other cheaper-substitute technologies.
Perhaps, if the "most widely deployed" claim were in any way even close to being true.
But it isn't. It's complete crap. MINIX's footprint in embedded controllers is still probably less than Linux's, and its overall footprint is dwarfed by Android.
I'm willing to wager there are more mobile phones in the world than intel ME enabled PCs at this point.
A quick, and entirely unverified, web search suggests Android devices (mostly not running on Intel ME-equipped CPUs) sell at a rate of around 1.4B per year, while Intel CPUs sell around 400M per year. So, yeah, MINIX is almost certainly way behind Android.
I'm not sure who first came up with that "most widely deployed" claim - I've seen it in a number of articles about ME and MINIX - but I doubt it's based on anything more than someone's wild-ass supposition.
Novell isn't really a company anymore, so much as a brand under Micro Focus (who bought Attachmate, who bought Novell).
Yes. As far as I know, that means we (Micro Focus) still own the copyright on the UNIX sources that Novell got from USL. In 2010, Attachmate confirmed that it intended to keep them after the merger; and while I don't remember seeing anything official, there are various stories online that refer to MF as "the current owner of UNIX".
Novell transferred the UNIX trademark and the rights to the UNIX specification to The Open Group, for use in the Single UNIX Specification.
Not very efficient. Simply announce a line of new Green Needles, a sustainable product made from dried herbaceous plant matter. Now you've reduced the problem to finding a needle in a needle stack.
Security is always last when implementing a new piece of software.
Not always. There are organizations that implement an SDLC correctly, with secure-development practices applied throughout the development process. And guess what? That leads to much lower security defect rates in new code, and gradually whittles away at security technical debt in old code.
Of course, the fact that your post (which, let's face it, is a cliche) was voted up to 5 Insightful shows that most places are still not implementing an SDLC, or they're doing it wrong.
He's a friend of Lady Mondegreen.