A comment must be made about the screen shots that are on display at One Terabyte of Kilobyte Age Tumblr now. At the moment, home pages from February 1997 are shown. But what one should remember when looking at the dates is that this is when they were updated for the last time. It is the date when a page was given up, not when it was made. What you see is not how the web looked in 1997, but how abandoned pages looked at that time!

For those wondering how the automated tumblr blog One Terabyte of Kilobyte Age Photo Op is made, the graphic below should be enlightening. More on technology, philosophy and historical value of screenshots will follow soon.

If you follow One Terabyte of Kilobyte Age Photo Op on tumblr, you’ve noticed many pages with broken images. You are not surprised, after all these pages are 17 years old right now. Decay fits to this age. But, important to notice, that “age” is not the only reason, or not the reason at all.

I examined the source code of pages with the broken images of the last 24 hours.


original url http://www.geocities.com/Colosseum/Field/2035/


original url http://www.geocities.com/Paris/LeftBank/1249/

null
original url http://www.geocities.com/Area51/3198/

null
original url http://www.geocities.com/SoHo/3162/

In cases above the reasons for images not being displayed is that they were included from external servers. Users “hot-linked” to files and services that, very probably, seized to exist even before Geocities went down.

In the cases below, users included images stored in the directory “pictures”, right under the root directory of the main Geocities server. How the <img> tags and their alt attributes appear in the HTML files gives away that this directory was used to store Geocities logos and other standard graphics like buttons and bullets. As it seems, none of them was saved during the rescue action of Archive Team. And Yahoo shut it down, though other directories with templates and backgrounds are still available. (Read more about ancient Geocities directories still accessible today in A First Sensation and The Ghost of Geocities.)

null
original url http://www.geocities.com/Hollywood/7990/

null
original url http://www.geocities.com/TheTropics/4884/

null
original url http://www.geocities.com/Hollywood/5197/

Many users spread their files over several accounts, for example the creator of Tokyo/4379 had another account for storing images at Tokyo/5261. The second directory was not archived (and doesn’t appear on reocitie’s list as well).
null
original url http://www.geocities.com/Tokyo/4379/

Here is an interesting example of a missing image — “INTRO.GIF”, as the source code states. The image is present at the reocities copy of the page, as “intro.jpg” It would be really interesting to know more about the reocities algorithm and approach to recovering pages.
null
original url http://www.geocities.com/Area51/6267/

There are many reasons for a webpage to lose its images, but in case of “Misty’s Home Page” there is no mystery. The author just linked all the images to his local hard disc. So they were never ever online! Well, to quote the webmaster: “My first attempt at having a home page […] As soon as I figure out how to do it, I will!”
null
original url http://www.geocities.com/WestHollywood/6152/

You can read more about digital ruins in my essay Ruins and Templates of Geocities.

Just got the book Netscape Communicator, published by Academic Press in 1998. “It presents a comprehensive strategy for total mastery of the Internet and of Web content creation.” I’ve skipped many pages and rushed immediately to “Adding Animated GIFs (p.548)”. It is a short chapter of only one and a half pages actually, saying that GIFs are great and as easy to add to your document as any other graphic. You should only know that not all of them are free for “unrestricted use”, but if you want to get free ones go to yahoo.com or … http://www.cswnet.com/~ozarksof/anigif1.htm

Chuck Poynter’s collection was obviously more prominent than I assumed before, and Dancing Girl is the mascot in its header. No wonder hulagirl.gif was spreading around amateur home pages so swiftly.

Read more about null in previous posts:


On the 4th of December 2012, Jason Scott, founder of Archive Team that restlessly rescues the files we uploaded all around the Internet, gave a wonderful lecture in Stuttgart. “Where are the Files?” tells the story of Jason’s personal obsession with copying and describes approachers and algorithms Archive Team uses to save the web. In some days the video version will be available at the Merz Akademie’s website.

In the end of his talk Jason confronted the audience with the three questions one should find answers to before signing up to an online service:

  1. What it can do that I can’t do myself?
  2. What are they doing with my stuff?
  3. Where is the “export” function?

The last issue is extremely interesting. Jason suggests that we shouldn’t deal with services which only offer upload options without providing reasonable ways to export your own data back to you. True! His next remark — that on the technical side implementing an “export button” is not much more work than to program an import one — also makes sense …

… but only if we assume that users want to download exactly the same what they have uploaded. That would mean if you uploaded 10 videos to youtube and then you clicked an export button and download them back to your hard disk. Or If you want just to take all your tweets, or photos, or songs back from where they were stored.

However, in my opinion, what users of Facebook, for example, would want to export is not the same as what they uploaded. They uploaded data, but want to get back the narration. To export a story is an effort on completely different scale, conceptually and programming wise.

Additionally, the story that you thought you were telling (“Timeline is your collection of the photos, stories, and experiences that tell your story“), is in fact told, designed, scripted, staged by the service. You can hit an export button till you die, it will not become yours.

So, back to the first question “What [a service] can do that I can’t do myself?”. To answer it you should probably know if you are able to narrate with your data yourself.


Some time ago Jason Scott posted an extremely beautiful 20 sec video “My Petabyte Roommate”. How comes it has only 300 views, I don’t know. I watch it 5 times every day since three weeks and can’t stop admiring the fact that computers that store web history appear like starry backgrounds of the earliest web pages.

CA: Yeah, so you’re surfing with a purpose, kind of. You have a hunch now that maybe people in the Pentagon neighborhood are using Felix, and you’re kind of also cataloging other information, but you’re mainly looking. Maybe it’s like these things are being presented to you, the information is being presented to you and you are just trying to identify what’s important.

(Cory Arcangel interviewed us for strictly-formal.org)

Every day, since March, after Philipp Budka’s presentation, I browse through profiles on MyKnet, Canada’s Aboriginal social network and can say that amount of people moving to Facebook is growing. At the same time, people with Facebook profiles continue to update their MyKnet profiles. Some times there are only reminders that the person has moved, some times proper illustrated reports about their life.

Some black screens:





I like the idea of A Book Apart to make “brief books for people who make websites”, and I order their books for my interface design students to provide them with text written by designers who are a) not making websites in Photoshop b) have their opinion about web design as a profession c) can write.

But the book #5 Designing for Emotion gives me hard time. Look at the titles of the first 4 chapters … and now imagine that they (and the rest of the book) are written without any mention or reflection of web sites people were making in the 90’s and are still making. Sites that are emotional and personal and human for real, not because of a marketing strategy.

The author, Aarron Walter, confesses in the beginning that for him the WWW started with dot.com with the gold zeros and gold ones rush. Maybe that’s why he doesn’t know or doesn’t want to know that the web outside of company websites and facebook profiles is and always was a very human environment. And beside it, makers of early web pages command an impressive set of instruments to appeal to the emotions of other users, to keep their audience and to be special and authentic.

The ignorance of web design professionals to the vernacular web started in 1997 and still progressing. Instead of valuing the rich history of their own medium, they’re proud when their web sites look like comics, magazine spreads, bottle labels or cartoons. Web designers, learn from the users, learn from the 19 (not 15!) years of web design, dig through the Geocities Torrent with us. It will bring you closer to “emotional design’s primary goal to facilitate human-to-human communication” (p.29), if this is really the goal.

P.S. Designers at Work

  1. Make a list of the files that need inspection with convmv, like this:
    (convmv --lowmem -r --nfc -f latin1 -t utf8 *) 2>&1 >> ~/Desktop/encoding_errors.txt
    You can find out here what files probably have filenames not encoded in utf8.
  2. Check out each directory containing bad filenames. Usually, inside one profile you will only encounter one encoding. To find out which one exactly, grep for parts of the bad file name in surrounding HTML files. For example:
    $ ls
    A?onet.jpg
    milenio.html
    $ grep 'onet\.jpg' *.html
    milenio.html:    <td colspan=3 rowspan=1 width=314 align="center" [rest of output line ommitted]

    Now you know that milenio.html links to the file that has a mysterious file name.

  3. Check out the HTML file in Firefox:
    $ firefox milenio.html
    By looking at the source code (Control+U) or “page information” (Control+I) you can most of the time guess the used encoding. Either

    • it is explicitly written into the <head> part of the HTML, like
      <meta http-equiv="content-type" content="text/html; charset=iso-8859-1" />
    • or the characters are written in entities inside the HTML, like &Agrave;, &#192; or %C0. Then the browser will usually display the correctly and you can try different encodings for the file.
    • If no encoding is specified anywhere, at least on Geocities the likelihood that it is iso-8859-1 is very high. (This used to be Netscape’s default encoding if no other was specified.)
  4. Use convmv to convert the file names:
    $ convmv -f iso-8859-1 -t utf8 -r --notest *
    The –notest option makes convmv actually do the renaming, if this option is omitted, convmv will just display what it would do to the files.