Thoughts on Maps

A few thoughts on Maps, in particular Google Maps and Apple Maps.

Google Maps was not the first digital maps solution nor even the first cloud maps solution. However, they have continuously refined the product adding information that they have gathered using high resolution satellite images, cars that roam the streets, and also local business information that they acquired through either partnering, web scraping, user-generated content and other means. Furthermore, based on the assumption that self-driving cars of the future will rely on maps that vastly more information intensive than the ones that we typically use today, having such high-quality maps is predicted by many to be a huge advantage when such technology becomes mainstream.

There are some problems with this approach though. One is sponsorship. Given that Google generates the vast majority of its revenues though advertising, it is highly probable that they will try to more aggressively monetise Maps in the not-so-distance future. This will most likely be through their integration of local business information, and by preferentially suggesting their sponsors’ businesses to their users, instead of simply listing in order of relevance and proximity. This may negatively impact the user experience in a way that has parallels to how Google has modified their Search results to preferentially show either their own services or sponsored products, relegating the more relevant organic search results to less convenient positions near the bottom of the page. It may also spark anti-trust concerns (and indeed already has) as this activity can be seen as an abuse of their dominant position on Maps to promote their own local business services.

Another is that the assumption that high-resolution maps are essential for autonomous driving will not last forever. If and when autonomous systems become as intelligent as human drivers, then they will no longer need such detail. They should be able to drive themselves based on the abstract and rudimentary maps that human drivers have relied on for decades. Therefore when AI becomes sufficiently sophisticated, these high-resolution maps will become mostly unnecessary and an overkill. The advantage of high-resolution maps will only last while autonomous systems are still much dumber than humans, and after that, their high costs of maintenance will make them more of a liability than an asset.

I point this out because there are apparently some who think that the current superiority of Google Maps compared to say Apple Maps provides an insurmountable moat for the future. While this may be true from a pure data volume and accuracy perspective, one should keep in mind that the disruptions that we have seen in the past decades are often those where the incumbent maintained an advantage in performance or sophistication, but nonetheless lost because the difference no longer mattered.