By John Gruber
Tuesday’s piece on Ninjawords was really about two stories. The small story is that of a clever $2 iPhone dictionary app, the developers of which removed “objectionable” words from its dictionary so as to get it published in the App Store. The big story is about the App Store itself, and whether Apple’s management is attempting to correct its course.
Yesterday afternoon I received a thoughtful email from Apple senior vice president Phil Schiller regarding Ninjawords and the App Store, and I think it bodes well for both stories. With Schiller’s permission, I’m reprinting the salient portions of it here:
When I read your column last night about the Ninjawords dictionary application I immediately investigated it with our App Store review team to learn the facts of what happened.
Let me start with the most important points - Apple did not censor the content in this developer’s application and Apple did not reject this developer’s application for including references to common swear words. You accused Apple of both in your story and the fact is that we did neither.
Ninjawords is an application which uses content from the Wiktionary.org online wiki-based dictionary to provide a nice fast dictionary application on the web and on the iPhone. Contrary to what you reported, the Ninjawords application was not rejected in the App Store review process for including common “swear” words. In fact anyone can easily see that Apple has previously approved other dictionary applications in the App Store that include all of the “swear” words that you gave as examples in your story.
The issue that the App Store reviewers did find with the Ninjawords application is that it provided access to other more vulgar terms than those found in traditional and common dictionaries, words that many reasonable people might find upsetting or objectionable. A quick search on Wiktionary.org easily turns up a number of offensive “urban slang” terms that you won’t find in popular dictionaries such as one that you referenced, the New Oxford American Dictionary included in Mac OS X. Apple rejected the initial submission of Ninjawords for this reason, provided the Ninjawords developer with information about some of the vulgar terms, and suggested to the developer that they resubmit the application for approval once parental controls were implemented on the iPhone.
The Ninjawords developer then decided to filter some offensive terms in the Ninjawords application and resubmit it for approval for distribution in the App Store before parental controls were implemented. Apple did not ask the developer to censor any content in Ninjawords, the developer decided to do that themselves in order to get to market faster. Even though the developer chose to censor some terms, there still remained enough vulgar terms that it required a parental control rating of 17+.
You are correct that the Ninjawords application should not have needed to be censored while also receiving a 17+ rating, but that was a result of the developers’ actions, not Apple’s. I believe that the Apple app review team’s original recommendation to the developer to submit the Ninjawords application, without censoring it, to the App Store once parental controls was implemented would have been the best course of action for all; Wiktionary.org is an open, ever-changing resource and filtering the content does not seem reasonable or necessary.
After going back to Ninjawords’s developers and conferring with some trusted sources within Apple, I believe what Schiller says here is genuinely the case — that what the App Store reviewers wanted for Ninjawords was a 17+ rating, not for Matchstick Software to filter its dictionary listings.
In particular, it seems like it really came down to bad timing around the launch of parental controls. Matchstick Software initially submitted the app on May 13. The response from the App Store was that Apple wouldn’t publish it with those words without a 17+ parental control rating. But parental controls — the preferences that specify the age rating limits for apps — debuted in iPhone OS 3.0, which was not released until June 17. And, it’s worth noting, the June 17 release date wasn’t announced until the WWDC keynote address on June 8. Back in May, Matchstick Software knew only that OS 3.0 was coming in the near future.
What Matchstick could have done was wait for iPhone OS 3.0 and publish the app with a 17+ rating. What they wanted to do, though, was ship their app as soon as possible. Hence Matchstick’s decision to begin filtering out the words which the App Store reviewers found objectionable. As Matchstick’s Phil Crosby told me via email last night, “17+ ratings were not available when we launched, which means at that time, it was simply not possible for our dictionary to be on the App Store without being censored. Given the options of censoring or sitting on the side lines while our competitors ate our lunch, we chose to launch.”
It is certainly arguable whether Wiktionary’s English dictionary content should be rated 17+. I personally disagree with that. But what I — and, judging from reaction around the web yesterday, many others — found outrageous was the idea that Apple insisted that Ninjawords both filter its dictionary and still carry the 17+ rating.
That is not the case, and that is good.
On the one hand, the 17+ rating for Ninjawords may not seem completely out of line, given that Apple is currently requiring the 17+ rating for many apps simply on the grounds that they display “unfiltered Internet content”. But Ninjawords does not display the live contents of Wiktionary — it contains a snapshot of the database. Unfiltered updates to Wiktionary do not make their way into Ninjawords over the air.
But more importantly, other dictionary apps in the App Store have innocuous age ratings, and yet contain all of the words that App Store reviewers objected to in Ninjawords (shit, fuck, and cunt, specifically). For example, American Heritage Dictionary ($30) is rated 9+; “Dictionary.com” is available from the App Store for free and is rated 4+.
(iPhone developers set their own ratings when they submit their applications to the App Store, but, as was the case with Ninjawords, the App Store reviewers will occasionally insist upon a certain minimum rating.)
Regarding this discrepancy between the ratings for dictionaries, Crosby said to me, “Apple may slap a 17+ rating on our app and wash their hands, saying ‘you’re not required to censor your app’, but at the same time, they’re putting a great deal of pressure on us to do so. Who wants to be the only illicit dictionary on the App Store? That may work for Urban Dictionary, but not us. I think that applying parental ratings inconsistently is tightly related to censorship in our case, and will be true for other apps as well.”
On the other hand, mDictionary — which, like Ninjawords, uses Wiktionary as its dictionary data source — is rated 17+. Nihongo, an English-to-Japanese dictionary app, is also rated 17+. Qingwen, a Chinese dictionary, is currently rated 4+, but its developer wrote a weblog entry about having been flagged by the App Store reviewers because the app contained Chinese translations for the words fuck, cock, and penis:
As is clear from the screenshots, Qingwen doesn’t bombard you with words like “cock” and “penis” the moment you start it up. No, the Apple employee who took those screenshots specifically searched for those words. As far as I’m concerned, it’s the same thing as opening a website that contains swear words (like the page you’re reading, for instance) on the iPhone. If they don’t want Qingwen on the iPhone because it can show you “objectionable material”, then why allow Safari, Mail, YouTube and pretty much any other app, which can easily show you all sorts of even more “objectionable material”?
So clearly Ninjawords is not alone in its treatment regarding profanity from App Store reviewers. I bought a copy of Qingwen to try the current version, and a query for “fuck”, the example flagged by the App Store reviewers, no longer generates any results. “Shit”, however, does.
Now, it’s possible that the Wiktionary database contains certain “urban slang” (to use Schiller’s term) that warrants a different rating from other English dictionaries, but the feedback Matchstick software received from the App Store reviewers was very specific. They were flagged for the words cunt, fuck, and shit. Those three words are in other dictionaries, rated as low as 4+. If there are other terms in the unfiltered Ninjawords dictionary that Apple considers more objectionable, why weren’t those terms pointed out?
I believe Phil Schiller that Apple’s policy is not to reject App Store dictionaries for containing swear words. However, it’s clear this policy has not been consistently enforced by the App Store review team. The problem, as I see it, is not that one or more App Store reviewers were unaware that it is acceptable for dictionaries to contain words that are not acceptable in other contexts. Mistakes are inevitable. The problem is that there’s no good recourse for developers to appeal such a mistake. It should have been enough for Matchstick Software to point out that the words flagged as objectionable in their initial rejection are in fact present in several other dictionaries already in the store.
Here’s how Schiller closed1 his email to me:
Apple’s goals remain aligned with customers and developers — to create an innovative applications platform on the iPhone and iPod touch and to assist many developers in making as much great software as possible for the iPhone App Store. While we may not always be perfect in our execution of that goal, our efforts are always made with the best intentions, and if we err we intend to learn and quickly improve.
This is music to my ears. That Schiller was willing to respond in such detail and length, on the record, is the first proof I’ve seen that Apple’s leadership is trying to make the course correction that many of us see as necessary for the long-term success of the platform. The improvement I consider most important is a significant focus on fairness, consistency, and common sense in the App Store review process.