I want to pose a question, as a prelude to looking at Amazon’s and Google’s quarters. The question is “Is the kind of intelligence portrayed in the movie ‘Her’ possible?” The reason for the question is that it’s clear that the notion of almost-human machine intelligence in the movie captured a lot of imaginations. It’s also clear that deep inside the plot line there are real technical paths to explore, real opportunities to address. For two companies whose future may depend on seizing the future, the question of “Her” feasibility is an important one.
Experts on the subject of machine intelligence, including one of Siri’s creators, are quick to point out that this movie pushes the boundaries significantly beyond the state of the art. I agree, but I think that misses the point. Can we make a machine so human it can provoke an emotion like love? Probably not, but we almost certainly can make one mimic human behavior quite effectively in a controlled context. To be sure, there are probably people who’d like to have a machine-intelligence friend/lover but most of us are just looking for machine-intelligence to be “intelligent”.
This is where Amazon and Apple and Google come in. Leaving aside those who want machine relationships and not intelligent answers, mobile broadband sets contextual boundaries on what people would look for from machine intelligence. I think I mentioned that the number one contextual question from a mobile user is “What’s that?” (closely followed by “Where am I?”) The combination of location, the specific visual focus, and the social context of a question considerably narrows the field of what’s expected, which makes it far easier to generate a machine response. You can imagine the rules associated with carrying on a conversation in abstract, then imagine the rules that are associated with someone walking down a street. In the latter context, the most likely conversations would be related to location/surroundings, so you could quickly tune voice recognition to preference this sort of inquiry. And if you could do this, you could create contextual mobile broadband services that people would buy.
According to Juniper Research, mobile users will spend $75 billion on apps by 2017. To put this into perspective, that’s more than the global telecom industry is expected to spend on IT in that same year (Ovum). How much money might be earned by supporting users in contextual services? I think, a lot, which is significant when you consider this in light of the recent numbers from Amazon and Google.
In Amazon’s case, you have a good story that wasn’t good enough; profits and revenues beat but not by as much as the Street had hoped for. What was most interesting about Amazon’s story in our view was that the cloud (AWS) didn’t really figure much into Street commentary. Why that’s notable is that if you pull out the notion of the cloud, then Amazon is nothing but a low-price online retailer and it’s really hard IMHO to justify the heady share price. I don’t think Amazon would be pushing the cloud if it didn’t think it had to, and yet the Street is focusing on its old-line retail business.
Amazon is in fact the largest and (currently) the most innovative player in the cloud space, but it’s just hitting growth targets and not blowing them away, despite the fact that it’s doing a lot in a technical sense to make its cloud better. The expanded model of platform services in the cloud is something Amazon has down perfectly, and exploits nicely with its Kindle. But Amazon needs to be a mass-market cloud player just like it needs to be a mass-market retailer, perhaps even more so. That means going beyond simply supporting Kindle to supporting the contextual service model that “Her” suggests is possible.
Some will say this argues for Amazon getting into the phone business, which many believe they will at some point. Actually fielding a smartphone and becoming an MVNO might in fact work for Amazon. The problem is that an Amazon phone might also be a major diversion at a time when Amazon doesn’t need one. The thing Amazon needs is a set of apps that can exploit AWS to mobile broadband users wanting contextual services. It doesn’t have to be linked to an Amazon phone; the profits would be better as a standalone and it would likely provoke a less dramatic competitive response.
Google has, perhaps, an even greater motivation. Google’s in-line results would very possibly have led their shares downward had it not been for the story that Google was looking to shed its Motorola handset business, whose margins and potential for growth the Street never liked. Google is now clearly counterpunching with Facebook in its positioning of mobile, and that is where I think they’re making a mistake.
At a high level, the Google decision (if proved) to shed mobile handsets is an indication that it’s not going to make mobile/Android directly profitable. Google is also admitting, with its new assertion that media is media and mobile is just one place to get it, that it’s not going to be able to lead in mobile advertising. Thus, they want to play down the value of the whole category. The problem is that, as I’ve been saying, online advertising is largely a zero-sum game and if Google takes a decisive step away from being a “mobile player” and asserts “mobile advertising” is just another face of the ad coin, it’s positioning away from some piece—perhaps the largest piece—of that zero-sum pie. That makes it a less-than-zero-sum for Google.
There is no player in the market who should have a better handle on contextual services than Google (except perhaps Apple, who appears equally clueless). They have all the pieces, including mapping, search, social networking chat and voice calling—there are few elements missing from that picture of context, except perhaps for the will to drive it forward. Given how the movie resonates with people for what networked intelligence and context could combine to do, maybe Google (or Amazon, or Apple) will finally get it. We can’t get to “Her” but we might get closer than anyone thinks.