[PREV - WWW]    [TOP]


I'm someone who typically looks askance
at GUI interfaces, preferring depth to
an often superficial "ease of use".

But consider command line ftp,
compared to the web.

Much of what's really interesting
about the web *could* have been         And some of the cooler things,
done with ftp sites                     like the Gutenberg Project,
                                        were being done that way
                                        in the pre-web days.
But look at the sheer number
of things you needed to do to                           (I can remember when
follow a reference:                                     the killer-app for
                                                        the internet was the
Reading through a document, you come across             sfraves calendar, sent
a statement that a certain file can be found            out via email.)
on an ftp site at a certain location, via
"anonymous" access.

(1)  You go into the ftp client:
     ftp some_machine.some_university.edu

(2)   It prompts you for user name, you answer "anonymous".

(3)   It prompts you for password, you type in your email address,

                                          (The first time you do this
                                          you puzzle over why it wants
                                          a password.)

(4)  If you're very lucky, you *know* the location, and do a
     cd <location>

  More likely, you need to do things like "ls" to
  see what choices you have, make guesses and cd
  up and down the tree looking for what you want.

                Quite often "ls" was incredibly
                slow... this was so bad that a customary
                kludge developed of putting a site map on
                the site intuitively named something like
                "ls-lrR.txt"  (because it was the output
                of a recursive ls command).

                    If you knew what you were doing, you
                    might transfer that one file first,
                    and look at it to figure out where you
                    should cd to.

(5)   Once you're in the right place, you just do something
      simple like "get <filename>".

(6)  Oh, wait a minute, where is the file going to be saved?
     Hopefully you remembered to change the *local* directory
     by doing something like "lcd /tmp".

        (It took me a long time to learn about that,
        because it has no analog among the standard
        unix commands, unlike "cd" and "ls".)

  Oh, and that thing you're downloading,
  it's just plain text isn't it?  Uh,
  no, actually it's *never* plain text,
  it's always compressed (frequently, a
  compressed tar archive).

(7)  So what you should have done, before
     starting the download, was to use the
     command "binary", to change modes and
     tell it you're planning on
     downloading something other than 7bit      (For reasons I won't
     ASCII text.  Otherwise, it mangles          get into, that
     the information.                            mangling was
                                                 considered a slick
                                                 feature, once).
    Modern ftp clients have the default
    fixed, but 'twas not so as late as the
    mid-90s, at least not in my experience

  So the odds are very good you're going
  to need to do the transfer again.

    Perhaps after being very puzzled for awhile about why
    that compressed archive isn't uncompressing.

(8)  Oh, and if that was something besides plain text you
     were after, e.g. a gif of an image, you were then going
     to have to feed the file name to a special viewer.

So that's very loosely around 8 steps...
And it's clunky and mistake prone enough
that you're going to need to re-do them
pretty often, so looking at something by
ftp might be something like a 12 step                "12 step"?
process on average.                                  No pun

   And then the web came along,                           Yet.
   and all of that was replaced
   by a single click.

   Suddenly you could browse a "docuverse" of sorts,
   without navigation getting in the way.

      Even a confirmed mouse-hater like
      myself had to admit that Mosaic had
      put the pieces in the right places...