iPhoto’s Lame-Oh Randomizer

   

Shot over 300 images over the Minnesota vacation, then whittled down to 120. The Achilles’ heel of digital photography is that there’s no risk/no expense, which encourages you to shoot five variants of everything, rather than one well-conceived shot. Nobody has any time, so the collections never get edited properly and you end up with mountains of superfluous bits to surf through in the future. With analog, each shot costs (financially, environmentally), so the image is conceived in the mind before being committed to film. Analog images are somehow less disposable.

It’s kind of like the difference between composing at the typewriter vs. the word processor (I wrote most of my college papers with a typewriter, only started using the UCSC mainframe during my senior year). When typing, mistakes are costly. So you roll your eyes, lick your lip, scratch your head, and conceive an entire paragraph mentally before committing to paper. Work from an outline so the pages come out in the right order. With word processing, you enter the process of infinite revision, spray your thoughts all over the page and let god sort ’em out (or do it yourself). Thoughts are more malleable with a word processor.

Anyway. Discovered last night that if you set iPhoto‘s slide show feature to randomize the images in an album, you’ll start seeing the same images over again very quickly.

– Displayed images are not dropped from the random queue
– The algorithm clearly favors some images, skipping others

Above: Miles at 11 months on the shores of Gull Lake, MN. Cousin Roya with famous goofy elastic mug.

Music: Etta James :: A Sunday Kind of Love

9 Replies to “iPhoto’s Lame-Oh Randomizer”

  1. >The algorithm clearly favors some images, skipping others

    No, it’s probably random. The thing is it’s vastly more likely that you’ll get sequences where some images repeat than a perfect random distribution. Think of coin flips: A billion heads-tails-heads-tails repetitions with no heads-heads-heads strings is extremely unlikely. Counterintuitive, but true.

  2. Heads-tails sure, but with 129 images in an album I would expect random to give me a pretty fair distribution. Or am I not understanding something there?

  3. Sure, if you let it go thru 50000 cycles you’ll have a pretty even distribution in the aggregate … but in any given small chunk of images you’re likely to have deviations from even.

  4. Yeah, but the aberrations here are more than just deviations. Like it will return to images #3 and #45 (e.g.) after every three images or something. I have no idea what they’ve done, but this is beyond what could be explained by deviates from pure average.

  5. Just underscores the point that you can’t get true stochastic outcomes from an algorithmic process. For true randomness, you need an outside source of entropy, like this:

    http://www.lavarnd.org/

    Still technically not random, because chaotic systems are still deterministic, but it’s effectively random (“epistemologically random?”), because you can’t measure the system’s initial state accurately enough to predict a future outcome.

    The only actually random phenomena we know about occur on the quantum level (radioactive decay), and even then it’s not clear whether it’s really random or whether it’s our knowledge that is imperfect. Philosophers hate this stuff.

    Excellent Ian Stewart article on the subj:
    http://www.fortunecity.com/emachines/e11/86/random.html

  6. There was a good piece in Wired recently (not online) about finding cheaper / more effective ways to generate true random seeds. One organization discovered a seemingly ideal solution: leave the lens cap on a video camera and monitor the white noise (black noise?) coming through the video feed. Run a hash on the aggregate or whatever to extract a truly random number. This can be done as effectively with a cheap tennis ball cam as with an expensive CCD system.

Leave a Reply

Your email address will not be published.