August 12, 2015 at 12:12 AM by Dr. Drang
You’ve probably had to explain to someone who isn’t into computers—a parent, a sibling, a spouse, a coworker—that computers do things the same way every time. That despite their belief that their computer has given them two different results even though they followed exactly the same steps each time, computers simply don’t work that way, and that they must have done something different the second time. We computer “experts” have faith in the deterministic nature of our machines. Web apps destroy that faith.
Apple Music is an example of this. A couple of weeks ago, Kirk McElhearn wrote a post about his experience with Apple Music making matches to your library based on metadata instead of the acoustic fingerprint of the song data itself. So if you have bad metadata in a song file, a different song entirely may be matched to it by Apple Music and it’s that different song that will show up on other devices linked to your account. Serenity Caldwell did her own testing on this and found that mismatches didn’t occur for her. Then Kirk redid his tests and found that the mismatches had gone away. Then Kirk re-redid his tests and found that the mismatches had returned.
Kirk and Serenity are obviously expert computer users, so what’s going on? Well, we can’t be absolutely sure, but it’s likely that Apple Music’s matching software—which is off running on servers far away—isn’t just one program. Different users in different areas are getting different versions of the software at different times as bugs are fixed (and introduced) in the “cloud.” Under these circumstances, it’s impossible to say what Apple Music will do because “Apple Music” is not one program; it’s several slightly different programs running on different servers around the world, with updates propagating slowly over the internet.
Or at least that’s one reasonable guess. There’s no way to know for sure because all the software that matters is remote from us. We don’t even know if we’re exercising the same software with each test.
I had a similar experience this evening. I decided to try out Twitter Cards, which can make links to your site look more attractive on Twitter. I added the necessary tags to my recent post about old graph paper and ran it through the Card Validator. The results looked good,
so I went ahead and tweeted out a link to it
Pardon the repetition. I’m testing Twitter cards.
— Dr. Drang (@drdrang) Aug 11 2015 5:56 PM
I looked at the tweet in the OS X Twitter app, the Twitter website on my Mac, and the iOS Twitter app on my iPhone. At first, none of them showed anything other than the tweet with the bare URL—no Card at all. But I heard from friends on Twitter that they were seeing the Card, and they sent screenshots to prove it.
Later, I found I could see the Card on both the website and in the Twitter app on my iPhone, but only when I clicked/tapped the tweet to see the more detailed view. Here’s what it looked like on the website:
Later still, the Card appeared in my home timeline in the iPhone Twitter app:
As I write this, the Card still doesn’t appear in the home timeline on the website until I click the “View summary” link. And, as you might expect, it doesn’t appear at all in the neglected OS X Twitter app.
None of them, you’ll notice, look like what the validator showed me: a full image view and all of the description text. I know there’s a reason for all these differences, but it sure seems like the behavior is random.
You’ve heard that old saying attributed to Einstein: “Insanity is doing the same thing over and over again and expecting different results”? With web apps, insanity is doing the same thing over and over again and expecting the same result.