So many little GUIs around these days, in mobile device apps and stuff – makes life feel a little more like navigating a video game

[me, phone in hand, leaving a building] “okay, now to press the button that shows me my position on the World Map”

argumate:

gasmaskaesthetic:

argumate:

gasmaskaesthetic:

I was trying to guess how many limes “five pounds of limes” is, so I pictured holding one lime, and thought “what’s a thing that feels about as dense as a lime and what’s five pounds of that” and I remembered my six pound Daschund that I had growing up, and she was probably a bit denser than a lime, so I tried to visualize how many limes it would take to be the size of my dog, minus her head, and came up with “thirty, but definitely between twenty and forty” and I was extremely close.  

does this count as a calibration exercise

I just can’t hold all these Dachshunds! 

would you rather fight one Daschund-sized lime or thirty lime-sized Daschunds

come on if you lose a fight against thirty lime-sized Dachshunds you die happy

hymneminium:

nostalgebraist:

Yesterday I re-read Neal Stephenson’s essay “In the Beginning was the Command Line.”  It’s just as entertaining as I had remember from reading it in high school, and also even more incoherent than I had thought at the time.

Much of it is just a charmingly besotted introduction to Unix/Linux culture and a primer on why it is so strange for operating systems to be big business, both of which are welcome, although probably old news to many.  The image with MS and Apple as car dealerships, and Linux as a nearby collection of yurts giving away tanks for free, is a great one.

But then there is all the stuff about Disneyland, “mediated experience,” text as a special medium, “metaphor shear,” and so forth, all of which is pretty much bullshit.  It is enjoyable bullshit, and perhaps there are non-bullshit ideas with the same general shape, but if one attempts to salvage this stuff, one crosses over the line from principle-of-charity contortionism into independent intellectual creativity.

---

Stephenson spends a lot of the essay riffing on the differences between GUIs and command line interfaces, which for him is a distinction of great philosophical weight.  He also talks about a supposed distinction between “text” and all other communicative media, of which the GUI / command-line distinction is a particular case.

The point of departure (let’s call this “Level 1″ in a taxonomy of Stephenson’s increasingly abstract riffs) seems to be the idea that you can use a GUI, but not a command line, without really knowing what you are doing.  With a command line utility, say, one can sometimes get pretty far without ever reading the man page, but one is always aware that one is using a tool built to precise specifications, and that its creators felt free to assume familiarity with these specifications.  But GUIs strive to be self-documenting, by always behaving in a way the user is presumed to find intuitive.

Now for what I’ll call “Level 2″: Stephenson elaborates on this by claiming that GUIs are metaphorical in a way that CLIs are not.  A CLI (supposedly) keeps you thinking about the work your computer is really doing, while a GUI presents a metaphorical world in which you (say) drag “documents” around on a virtual “desktop,” and encourages you to pretend you are really interacting with this metaphorical world, not giving commands to a machine.

Finally, in “Level 3,” Stephenson claims that this distinction (or something like it) differentiates text itself from all other media.  This allows for a unified elitism in which “people who read books” are spiritual kin of Unix hackers, a priestly caste (literally “Morlocks” as opposed to “Eloi”!) who engage with text and deeply understand reality.  My goal isn’t to criticize Stephenson for being elitist, though, just for being incoherent.

The moment in the essay where Stephenson makes the “text is special” claim is a remarkable bit of misdirection.  Here it is:

Among Hollywood writers, Disney has the reputation of being a real wicked stepmother. It’s not hard to see why. Disney is in the business of putting out a product of seamless illusion–a magic mirror that reflects the world back better than it really is. But a writer is literally talking to his or her readers, not just creating an ambience or presenting them with something to look at; and just as the command-line interface opens a much more direct and explicit channel from user to machine than the GUI, so it is with words, writer, and reader.

The word, in the end, is the only system of encoding thoughts–the only medium–that is not fungible, that refuses to dissolve in the devouring torrent of electronic media (the richer tourists at Disney World wear t-shirts printed with the names of famous designers, because designs themselves can be bootlegged easily and with impunity. The only way to make clothing that cannot be legally bootlegged is to print copyrighted and trademarked words on it; once you have taken that step, the clothing itself doesn’t really matter, and so a t-shirt is as good as anything else. T-shirts with expensive words on them are now the insignia of the upper class. T-shirts with cheap words, or no words at all, are for the commoners).

Note, first, that that huge parenthetical is a tangent.  Even if it were true (which it isn’t), it does not constitute an argument for the claim before the parenthetical.  But it successfully distracted me, so that it was only after reading this several times that I noticed there is no argument for the claim.  Excising the parenthetical, we simply have:

The word, in the end, is the only system of encoding thoughts–the only medium–that is not fungible, that refuses to dissolve in the devouring torrent of electronic media[…].

This simultaneously florid and vague assertion (“refuses to dissolve in the devouring torrent”?) cries out for explication, but in the very next paragraph, Stephenson proceeds as though it has been elaborated and firmly established:

But this special quality of words and of written communication would have the same effect on Disney’s product as spray-painted graffiti on a magic mirror. So Disney does most of its communication without resorting to words […]

And on he goes, for the rest of the essay.

By the way, that parenthetical is nonsense.  First, copyrights are rarely used in fashion.  Second, trademarks are frequently used in fashion, but images can be trademarked just as well as words can, and they frequently are.  The idea that logos can be easily bootlegged, but text cannot be due to some inherent quality of text, seems like a non-starter.  (It is possible to imagine a contingent legal situation in which text could be trademarked but images could not be, but even this is not the case.)

---

What about Level 2?  Stephenson is worried about “metaphor shear,” a term which seems approximately equivalent to “abstraction leakage.”  For instance:

Anyone who uses a word processor for very long inevitably has the experience of putting hours of work into a long document and then losing it because the computer crashes or the power goes out. Until the moment that it disappears from the screen, the document seems every bit as solid and real as if it had been typed out in ink on paper. But in the next moment, without warning, it is completely and irretrievably gone, as if it had never existed. The user is left with a feeling of disorientation (to say nothing of annoyance) stemming from a kind of metaphor shear–you realize that you’ve been living and thinking inside of a metaphor that is essentially bogus.

I feel like this is almost too obvious to point out, but almost all useful abstractions leak sometimes, and if there is a bottom level at which abstractions stop being leaky, it is way below using a CLI.  When using a CLI, you let yourself be “lulled into the metaphor” (if we must speak this way) that you are interacting with an abstract system that accepts letters from you, composed into lines which are executed.  You don’t know, and don’t have to know, how the letters you see are really represented in computer memory, or how it is that the computer ensures the letters you type are represented stably on the screen.

And of course (again, pointing out the obvious) all of this software stuff is an abstraction on top of hardware, which is maintained as a stable illusion for users who don’t have to think about CPU registers and caches or VLSI, much less semiconductor physics.  Stephenson rhapsodizes about C and compilers, without noticing that C is a leaky abstraction on top of byte code, and that you can easily use gcc without understanding how it works.

It is deeply strange to me that someone presenting himself as conversant with the software world would field this sort of argument.  Like, I’m not all that conversant with the software world, but even I know that “abstractions that let you ignore lower-level stuff most of the time” are the only way anything ever gets done, ever, period.

---

To be fair, there is a real difference when you start talking about abstractions specifically intended for people who don’t know the lower-level stuff and don’t want to.  (This is the “Level 1” distinction.)  However, this has little to do with text vs. other media, or with metaphor vs. reality.

What makes GUIs (and relatives) special here is that one must be able to sit down with them and immediately get intuitively expected behavior out, without needing to first get a handle on precisely what function is being computed (so to speak), what its inputs and outputs are.  Stephenson talks about the power of the Unix find command.  I’ve been using Unix CLIs for many years now, albeit not as a power user, but I still don’t fully understand the “find” command.  Usually, if I have to find something on my Mac, I can find it using the GUI’s search bar, which is set up to do exactly the thing I want 99% of the time, the moment I type something in.  I don’t know how configurable it is beyond that, and I don’t especially care.

This is not a GUI vs. CLI distinction; one can imagine a CLI tool that does the exact same thing as the Mac search bar (and this almost certainly exists).  The find command is the way it is because (among other things) it is meant to be piped into/out-of other Unix commands and all that good stuff.  But one has to trade off other desiderata to get that property, which is frequently not needed anyway.  This has nothing to do with text, Disneyland, or any of this fancy stuff.  It’s just using the right tool for the job.

---

A particular irony here is that Stephenson’s essay, an ode to the precision of text (CLIs force you to be precise), is itself a piece of imprecise text.  It is full of leaky metaphors, and papers over technicalities with the writerly equivalent of fancy graphics.  Reading it feels like using an overstuffed GUI.

Stephenson’s view makes more sense if you apply it to traditional Unix CLIs, instead of CLIs in general. In traditional Unix, each program only does one thing. cat just outputs the files you give it as arguments. It’s hard to use cat without understanding what it does, because it’s so simple. There’s a beauty to that kind of minimalism.

But not all CLIs are like that. find is distinctly non-Unixy (even though it’s part of Unix). Linux and GNU reject the minimalism. GNU’s cat can do line numbering and convert non-printing characters. GNU ls can give different output depending on whether it’s talking to a terminal or not. Some of the old style is left, because it has to be compatible with the old tools, but you’re never aware of what the tools do exactly, unless you look inside and read the source code.

fish bravely attempts to have all the friendliness and discoverability of a GUI, implemented in a CLI. I think I still don’t exactly understand all the details of the strategies behind its autocompletion, but it consistently works really well, so I don’t need to know.

And a few GUIs do manage the traditional Unix minimalism. Something like slock is simple enough to understand completely, despite being graphical.

Stephenson describes a real difference, but it’s a difference that’s only loosely tied to CLIs and GUIs.

It’s not about CLIs, it’s about programs with simple behavior, simple implementations, and an external interface that’s tightly coupled to what’s going on inside.

It’s not about GUIs, it’s about programs that try to make it easy for the user to figure out how to do what they want, and don’t match their interface to their implementation.

Many CLIs are in the second category. A few GUIs are in the first.

Yesterday I re-read Neal Stephenson’s essay “In the Beginning was the Command Line.”  It’s just as entertaining as I had remember from reading it in high school, and also even more incoherent than I had thought at the time.

Much of it is just a charmingly besotted introduction to Unix/Linux culture and a primer on why it is so strange for operating systems to be big business, both of which are welcome, although probably old news to many.  The image with MS and Apple as car dealerships, and Linux as a nearby collection of yurts giving away tanks for free, is a great one.

But then there is all the stuff about Disneyland, “mediated experience,” text as a special medium, “metaphor shear,” and so forth, all of which is pretty much bullshit.  It is enjoyable bullshit, and perhaps there are non-bullshit ideas with the same general shape, but if one attempts to salvage this stuff, one crosses over the line from principle-of-charity contortionism into independent intellectual creativity.

Stephenson spends a lot of the essay riffing on the differences between GUIs and command line interfaces, which for him is a distinction of great philosophical weight.  He also talks about a supposed distinction between “text” and all other communicative media, of which the GUI / command-line distinction is a particular case.

The point of departure (let’s call this “Level 1″ in a taxonomy of Stephenson’s increasingly abstract riffs) seems to be the idea that you can use a GUI, but not a command line, without really knowing what you are doing.  With a command line utility, say, one can sometimes get pretty far without ever reading the man page, but one is always aware that one is using a tool built to precise specifications, and that its creators felt free to assume familiarity with these specifications.  But GUIs strive to be self-documenting, by always behaving in a way the user is presumed to find intuitive.

Now for what I’ll call “Level 2″: Stephenson elaborates on this by claiming that GUIs are metaphorical in a way that CLIs are not.  A CLI (supposedly) keeps you thinking about the work your computer is really doing, while a GUI presents a metaphorical world in which you (say) drag “documents” around on a virtual “desktop,” and encourages you to pretend you are really interacting with this metaphorical world, not giving commands to a machine.

Finally, in “Level 3,” Stephenson claims that this distinction (or something like it) differentiates text itself from all other media.  This allows for a unified elitism in which “people who read books” are spiritual kin of Unix hackers, a priestly caste (literally “Morlocks” as opposed to “Eloi”!) who engage with text and deeply understand reality.  My goal isn’t to criticize Stephenson for being elitist, though, just for being incoherent.

The moment in the essay where Stephenson makes the “text is special” claim is a remarkable bit of misdirection.  Here it is:

Among Hollywood writers, Disney has the reputation of being a real wicked stepmother. It’s not hard to see why. Disney is in the business of putting out a product of seamless illusion–a magic mirror that reflects the world back better than it really is. But a writer is literally talking to his or her readers, not just creating an ambience or presenting them with something to look at; and just as the command-line interface opens a much more direct and explicit channel from user to machine than the GUI, so it is with words, writer, and reader.

The word, in the end, is the only system of encoding thoughts–the only medium–that is not fungible, that refuses to dissolve in the devouring torrent of electronic media (the richer tourists at Disney World wear t-shirts printed with the names of famous designers, because designs themselves can be bootlegged easily and with impunity. The only way to make clothing that cannot be legally bootlegged is to print copyrighted and trademarked words on it; once you have taken that step, the clothing itself doesn’t really matter, and so a t-shirt is as good as anything else. T-shirts with expensive words on them are now the insignia of the upper class. T-shirts with cheap words, or no words at all, are for the commoners).

Note, first, that that huge parenthetical is a tangent.  Even if it were true (which it isn’t), it does not constitute an argument for the claim before the parenthetical.  But it successfully distracted me, so that it was only after reading this several times that I noticed there is no argument for the claim.  Excising the parenthetical, we simply have:

The word, in the end, is the only system of encoding thoughts–the only medium–that is not fungible, that refuses to dissolve in the devouring torrent of electronic media[…].

This simultaneously florid and vague assertion (“refuses to dissolve in the devouring torrent”?) cries out for explication, but in the very next paragraph, Stephenson proceeds as though it has been elaborated and firmly established:

But this special quality of words and of written communication would have the same effect on Disney’s product as spray-painted graffiti on a magic mirror. So Disney does most of its communication without resorting to words […]

And on he goes, for the rest of the essay.

By the way, that parenthetical is nonsense.  First, copyrights are rarely used in fashion.  Second, trademarks are frequently used in fashion, but images can be trademarked just as well as words can, and they frequently are.  The idea that logos can be easily bootlegged, but text cannot be due to some inherent quality of text, seems like a non-starter.  (It is possible to imagine a contingent legal situation in which text could be trademarked but images could not be, but even this is not the case.)

What about Level 2?  Stephenson is worried about “metaphor shear,” a term which seems approximately equivalent to “abstraction leakage.”  For instance:

Anyone who uses a word processor for very long inevitably has the experience of putting hours of work into a long document and then losing it because the computer crashes or the power goes out. Until the moment that it disappears from the screen, the document seems every bit as solid and real as if it had been typed out in ink on paper. But in the next moment, without warning, it is completely and irretrievably gone, as if it had never existed. The user is left with a feeling of disorientation (to say nothing of annoyance) stemming from a kind of metaphor shear–you realize that you’ve been living and thinking inside of a metaphor that is essentially bogus.

I feel like this is almost too obvious to point out, but almost all useful abstractions leak sometimes, and if there is a bottom level at which abstractions stop being leaky, it is way below using a CLI.  When using a CLI, you let yourself be “lulled into the metaphor” (if we must speak this way) that you are interacting with an abstract system that accepts letters from you, composed into lines which are executed.  You don’t know, and don’t have to know, how the letters you see are really represented in computer memory, or how it is that the computer ensures the letters you type are represented stably on the screen.

And of course (again, pointing out the obvious) all of this software stuff is an abstraction on top of hardware, which is maintained as a stable illusion for users who don’t have to think about CPU registers and caches or VLSI, much less semiconductor physics.  Stephenson rhapsodizes about C and compilers, without noticing that C is a leaky abstraction on top of byte code, and that you can easily use gcc without understanding how it works.

It is deeply strange to me that someone presenting himself as conversant with the software world would field this sort of argument.  Like, I’m not all that conversant with the software world, but even I know that “abstractions that let you ignore lower-level stuff most of the time” are the only way anything ever gets done, ever, period.

To be fair, there is a real difference when you start talking about abstractions specifically intended for people who don’t know the lower-level stuff and don’t want to.  (This is the “Level 1” distinction.)  However, this has little to do with text vs. other media, or with metaphor vs. reality.

What makes GUIs (and relatives) special here is that one must be able to sit down with them and immediately get intuitively expected behavior out, without needing to first get a handle on precisely what function is being computed (so to speak), what its inputs and outputs are.  Stephenson talks about the power of the Unix find command.  I’ve been using Unix CLIs for many years now, albeit not as a power user, but I still don’t fully understand the “find” command.  Usually, if I have to find something on my Mac, I can find it using the GUI’s search bar, which is set up to do exactly the thing I want 99% of the time, the moment I type something in.  I don’t know how configurable it is beyond that, and I don’t especially care.

This is not a GUI vs. CLI distinction; one can imagine a CLI tool that does the exact same thing as the Mac search bar (and this almost certainly exists).  The find command is the way it is because (among other things) it is meant to be piped into/out-of other Unix commands and all that good stuff.  But one has to trade off other desiderata to get that property, which is frequently not needed anyway.  This has nothing to do with text, Disneyland, or any of this fancy stuff.  It’s just using the right tool for the job.

A particular irony here is that Stephenson’s essay, an ode to the precision of text (CLIs force you to be precise), is itself a piece of imprecise text.  It is full of leaky metaphors, and papers over technicalities with the writerly equivalent of fancy graphics.  Reading it feels like using an overstuffed GUI.

did you mean “X > Y” instead of “X < Y”?

I think it’s correct as written.  It’s supposed to sound intuitively wrong.  (It’s from a paper that says any formal population ethics system has to imply one of 3 intuitively wrong conclusions.)