[michaelt]

  The algorithm examines 42 
  different signals, they said, 
  including an app’s relevance
  to a given search, its
  ratings, and its popularity
  based on downloads and user
  clicks.
  [...]
  If you searched for “podcast”
  in May 2018, you would have
  had to scroll through as many
  as 14 Apple apps before
  finding one made by another
  publisher. 
  [...]
  “We make mistakes all the 
  time,” Mr. Cue said.
  “We’re happy to admit when we
  do,” Mr. Schiller said. “This
  wasn’t a mistake.” 

I have to say, it stretches credulity to claim any non-faulty algorithm would put Apple's "compass" app as the second result in a search for "podcast"

It's difficult to make any sense at all of such a result.

[OJFord]

Ranking by count the all apps installed by users who have previously searched for <term> might do it.

Not saying that's a good algorithm, but one _might_ be tempted to think that with enough users it should be fine. The 'bug' being that all Apple's users are Apple users, so Apple apps would get pushed high for any search.

I'm certainly not suggesting that's what happeened (or even that it would be something so simplistic), but it gives me an idea at least how something that, when presented only with the symptoms, seems so unbelievable.

[HelloMcFly]

Agreed, but the end result is that their algorithm pre-fix was clearly faulty and should be considered a mistake. Their gymnastics to the contrary just come across as either a bad-faith argument or self-delusion.

[Bud]

What comes across as cherry-picking to me is pointing out a single random inconsequential web search and the result it used to have in a previous year.

[yyyk]

Not a single search, but many searches done repeatedly over time, proving the result is highly unlikely to be happenstance. Oh, and the results only happened to changed after Spotify launched an official complaint. The WSJ article the NYT links to is even more detailed.

[Wowfunhappy]

Shouldn't "Facebook" also turn up near the top, in that case?

[bobwaycott]

For those on mobile/small screens, a better formatted quote:

> The algorithm examines 42 different signals, they said, including an app’s relevance to a given search, its ratings, and its popularity based on downloads and user clicks. ...

> If you searched for “podcast” in May 2018, you would have had to scroll through as many as 14 Apple apps before finding one made by another publisher. ...

> “We make mistakes all the time,” Mr. Cue said.

> “We’re happy to admit when we do,” Mr. Schiller said. “This wasn’t a mistake.”

For those on desktop, please stop formatting quotes like code. They’re just impossible to read on anything but a very wide screen.

[JustSomeNobody]

What gets me is that if you do it now, Overcast is about 9th and Pocket Casts is 25th or something. Those are two very popular podcast apps. I stopped scrolling looking for Castro.

[toyg]

42 different signals, really? Smacks of trolling: https://en.wikipedia.org/wiki/42_(number)