PaulFarhi
Latest
Siri VC Morgenthaler weighs in on iPhone 4S bandwidth badmouthing (Updated)
Back in early January, wireless optimization firm Arieso released a report on the bandwidth usage profiles of various smartphones and other devices. It noted a surge in download and upload usage for the iPhone 4S, moving the new phone ahead of the iPhone 4 and various Android devices to take over the top spot for smartphone bandwidth consumption. At the time, coverage by Reuters and Bloomberg (reiterated by several sites, sorry to say, including TUAW) put the onus for this bump in the pipe on one of the marquee features of the 4S: the voice-controlled digital assistant, Siri. [Arieso's CTO Michael Flanagan spoke about Siri's role in bandwidth usage in a video interview with Bloomberg back when the report was released, and while he doesn't exactly pin all the blame on Siri -- he says the usage is due to increased utilization of 'cloud services' -- Siri does come in for some of the blame. It's not nearly as emphatic as subsequent reports would have it. Video embedded below. –Ed.] Fast forward a few weeks: an op-ed from Paul Farhi in the Washington Post dives back into the Arieso report, recycling the charge that Siri is singlehandedly responsible for pumping up the load on America's cellular networks and degrading data service for everyone. (Farhi stopped short of blaming her for fluoride in the water and the rise of reality television.) This time, however, some consideration of the facts in evidence led us to question his (and our) original conclusions; we know that the math around Siri's data usage just doesn't add up to a consequential amount. We also reached out to Arieso for a copy of the full report, which the company happily and promptly provided. Now there's another voice weighing in, and it belongs to someone who's in a position to know what's what: former Siri board member Gary Morgenthaler. He wrote a guest post for Forbes's web site that thoroughly debunks the notion of Siri as a bandwidth bandit. Siri actually uses only small bites of wireless data, as tested by Ars Technica and backed up by Morgenthaler's inside info. What does Morgenthaler suspect might be at play in the 4S bandwidth numbers? He's careful to couch it as speculation rather than assert it as fact, but the addition of iCloud to the iOS service mix is a likely culprit (especially since buyers of new phones are more likely to set up Apple's cloud service than users of older models, who may be slower to update to iOS 5). Other possibilities include iTunes Match, Photo Stream, or simply the wider chunk of image data captured by the iPhone 4S's improved camera. As Mel noted earlier, the 4S also supports faster 3G downloads than the earlier models. [Photo Stream is WiFi-only, as pointed out by David Barnard.] There's only one issue I saw in Morgenthaler's response, but it's an important one. Morgenthaler suggests that Arieso committed a logical error in citing Siri as the cause of the 4S results: Arieso presents no data to support their claim that Siri causes the increase in iPhone 4S data usage. Most likely, Arieso committed a logical fallacy which, in grammar school, was called "post hoc, ergo propter hoc" - that is, "after the fact, therefore, because of the fact." In other words, Siri is new and associated with the iPhone 4S; therefore, it must be the cause of this increased traffic. It's an appealing argument, but it doesn't hold water. That's also an appealing argument, but he may be pinning the propter hoc on the wrong donkey. Even a cursory review of the Arieso report is sufficient to discover that it does not mention Siri at all, and it makes no such claim that the voice assistant is responsible for the bandwidth hunger -- at least, not in the body of the report. When it comes to the cause of the iPhone 4S's download desire, in fact, the report makes it very clear that it's too soon to draw any specific conclusions (emphasis mine): The iPhone 4S showed an increase of 176% in downlink data volumes over the iPhone 3G. Since the downlink-to-uplink data volume ratio was almost 7-to-1 on average for the devices under study, this downlink increase of 176% corresponds to a larger total volume of data than a 220% uplink increase (discussed in the last section). As noted earlier regarding the increases in total numbers of data calls, it remains a topic for further study to characterise the root cause of this downlink data volume increase. Arieso's research never said Siri was completely to blame, at least not in so many words; it's not clear whether a company representative said something to Reuters to encourage this conclusion in CTO Michael Flanagan's interviews he does say that Siri is a potential part of the issue, but that conclusion is clearly not backed up by the report itself. Did the firm make an effort to correct the record, or was there a calculation that the Siri-related media attention would be more valuable than getting the accurate information out? We don't know, but we've reached out to Arieso's PR folk for comment. It begins to look less like an error of logic and more like a calculus of publicity. Update: Shortly after this post was published, I did speak with Flanagan about the report and the surrounding controversy. While he acknowledges that some of the media coverage of the bandwidth report may have weighed overmuch on Siri, he disavowed any conclusions not present in the original report (despite the fact that the report was only available on request, and as such most interested parties would not have read it). As noted above, the report specifically does not point any fingers regarding the cause of the bandwidth bump. Flanagan did allow that perhaps Arieso will be a bit more assertive about correcting unfounded assumptions the next time the company issues a bandwidth usage report. The Next Web also picked up a key tidbit from the research: the iPhone 4 utilization numbers were from 2010 (rather than comparing the 4 and the 4S during the same time period), meaning that the overall landscape may have shifted during the intervening months. Meanwhile, we're left with the unavoidable conclusion: The Washington Post and Paul Farhi picked a fight with Siri for something she does not do, picking up the "Siri is a data guzzler" concept from the original coverage -- but not from the actual data. Hanging the 4S bandwidth bulge on an innocent intelligent assistant may be great for headlines, but it looks to be wrong on the facts.
Siri probably isn't the bandwidth hog the WaPo warned you about
Update: Former Siri director responds to the Washington Post story. Reuters was the original source of the error around Siri bandwidth usage. "How Siri is ruining your cellphone service." That's the searing headline from the Washington Post in an article by Paul Farhi. Farhi claims that "Siri's dirty little secret is that she's a bandwidth guzzler, the digital equivalent of a 10-miles-per-gallon Hummer H1." Where's he coming up with this? Apparently, the "Siri eats bandwidth" claim is based on a study by Arieso that reports that iPhone 4S owners consume twice as much cellular data as iPhone 4 users and 3 times as much as iPhone 3G users. Recent Android phones are also chewing up twice as much data as the iPhone 3G, while 3G and 4G mobile hotspots are by far the biggest download hogs (26x the baseline). At least in the press release summary of the study, however, there's no mention of Siri at all; just the increased usage for the 4S, which just happens to support a faster download standard on AT&T's network. Our sister site Engadget helped put that study in perspective by pointing out that Arieso has a vested interest in the results of the research. We've asked for a full copy of the report to see what, if any, linkage there is between Siri and data volume. [Ars Technica did a round of testing when the iPhone 4S came out, and the results showed that an average Siri user might add just 10MB of data consumption per month due to voice queries. Considering that the high-end AT&T data plan provides two gigabytes of traffic, that 10MB would represent only 0.5% of the full allocation, or 1% of a half-used allowance -- unlikely in the extreme to result in a doubling of data usage. –Ed.] If we take the study at face value, though, why more data on the 4S? The likely answer hasn't much to do with Siri and a lot more to do with the profile of the iPhone 4S buyer. The people who buy the latest phone are also the power users who take the most advantage of their devices. We've seen that happen before with new technology, and once people stop amazing themselves and their friends, the consumption of bandwidth drops off. I haven't seen any convincing data that says the iPhone 4S inherently uses more data than an iPhone 4, and iOS 5 iCloud features, also available on the iPhone 4 and 3GS, probably play a role in increased bandwidth use. As for Siri, most of the heavy lifting goes on at the Apple servers, where your query is translated into data and then sent back to your phone in a quick burst. Streaming radio, Netflix and a host of other apps can use way more bandwidth, and they are utilizing the network for minutes or hours at a time, not seconds. Of course Siri is on every iPhone 4S, so it is getting used more than some 3rd-party apps, but it's hard to believe that the average user doing perhaps 2-3 queries a day is destroying our cellular infrastructure. GigaOm this morning also poured cold water on the Post story, and there will probably be more to come. One thing is for sure. Smartphones, and the iPhone in particular, are using more data than the dumb phones of old. Compare that to the internet connections in our homes, where Netflix has been identified as the biggest user of bandwidth in the U.S. It's up the internet providers, both wired and wireless, to keep growing their networks so they can continue to charge those premium rates; it's also up to Washington regulators and cellular carriers to make efficient use of bandwidth and future spectrum technologies. Readers, are you heavy Siri users, and are you destroying our cellular networks?
