Google’s New Patent – Capturing The Data

At the heart of the Google method is capturing and interpreting data from an array of sensors on mobile devices as well as from related behavior of mobile users and other sources. These data sources, according to the patent, include:

Sensors on Smartphones: Microphone, Orientation Sensor (that “can sense a user’s gestures”), Gravitometer, Accelerometer, Camera, Display, Barometer, On/Off Condition, Ambient Charge Sensor, and Temperature Sensor.

Other Sources: Historic user-specific data, Historic location data, Current neighboring micro-climate data, Current weather forecasts, and Crowdsourced reports of actual weather.

Note that the mobile sensor data depends on there being mobile device users in the exact area (or at least nearby) at the time that the system user makes their inquiry about the current weather that will affect him/her immediately. The patent notes that if there aren’t a reasonable number of devices present, the sensor data may not be reliable.

Google’s Imaginative Uses of Data

Google basically appears to be trying to squeeze every possible shred of meaning from all of the abundant sources of data that it lists.

All of sensor data is fed into a “micro-climate manager.” The patent states, “Generally, micro-climate manager is capable of determining, based on current weather-related sensor data, a current or projected micro-climate for a micro-location.”

Google contemplates that the actual micro-climate data will be transferable to an individual on virtually any type of device such as a mobile phone, TV screen, wearable, computer, or even a home appliance. It refers to these devices as “computing devices.”

Regarding use of non-sensor data, the patent gives an example where there are very few people (e.g., 6) on the street the user wishes to walk on, but many people (e.g., 48) on a parallel street. The data from the six phones is considered inadequate, so the system may use the weather reported from the next street. But in that case it will consult historic records to determine the correlation between the weather on the two streets.

The patent goes on to state that the system may even look to a far wider area, even a street two miles away if the distant street “has a strong correlation” historically for the weather condition in question (such as high winds at that time of day) with the user’s specific street.

In addition, the system, i.e., the micro-climate manager, may look to a forecast of weather for the entire area or region, if it feels that is likely to be accurate for the micro location in question.

Finally, note, that the manager may also defer to actual crowdsourced data, e.g., a user who is on the street in question who reports in the weather conditions at the moment.

It’s obvious that the company is relying on a good deal of AI and Big Data processing, to be able to analyze all of these disparate sources and to come up with immediate responses to user requests.

Our Take: Google’s Ambitious Scope

The typical examples cited in the patent materials are seemingly simple and mundane. User X is about to leave his apartment in Manhattan and walk to the subway a few blocks away. The system has gathered weather condition data from multiple sensors on dozens of smartphones that are on the streets in questions at that time.

The micro-climate manager digests the sensor inputs and can tell the user how cool, windy, wet, etc., the conditions are he is going to encounter. The manager is also assumed to know the person’s normal route for walking to the subway and may suggest an alternative route to avoid more adverse weather conditions on the customary path.

This straightforward example raises a number of questions about whether it would be worthwhile to Google to attack this complex undertaking, even if it could greatly improve millions of people’s confidence about the weather they are going to encounter. After all, there are so many caveats implied in this patent: Primarily – if there aren’t enough smartphones out there on the streets you’re interested in, the best you are going to get is a judgment based on some AI applied to historic data, or even just the local weatherman’s forecast.

This alone means that the system’s coverage, in our opinion, might be fairly strong on the downtown streets of major metropolises, but not so great elsewhere.

And there are other sources of hyper-local weather forecasts on the market. For example, Dark Sky claims that it can:

“…predict when it will rain or snow, down to the minute, at your exact location: we deliver hyper-local forecasts, not just for your city or state, but right where you’re standing.”

However, when we look beyond the simple typical example cited in the patent, what Google is suggesting could be extremely ambitious. It indicates that it will assemble mammoth amounts of data about weather for the micro locations it chooses to cover.

Leaving aside the issue of whether this will provide an instant, accurate slice of information to one specific user at a give moment, the data may be very helpful for longer term uses in understanding weather and weather risk for many purposes.

Insurance provider Allianz issued a study that pegged the losses in the U.S. alone at about 3.4% of GDP (about $600 billion) annually and suggested that a large part of this could be avoided with better data and preparatory tools and procedures. This study only looked at the risk side of the equation; with better understanding of micro weather, there would also come new opportunities to enhance business operations.

Back in April 2016 we wrote about the huge business potential in Mobile Weather, stating:

“The next five years will see a massive proliferation of new weather-related apps, and analytics.” (“Mobile Weather’s Big Growth Potential,” MCE 4/18/16)

We’ve been waiting for some time to see Google capitalize on Weather as an area that can be key to its overall strategies. We view Google as a company that has set up a platform for trying to enmesh users in its services as many minutes of the day as possible. Weather is not only the largest area of Big Data, but it is a factor for everybody every day.

Four years ago we spoke with an executive at Microsoft who told us he believed Google would acquire The Weather Channel. Instead key parts were bought by IBM, which does not have the platform to truly exploit weather the way Google could. We will be following closely to see how vigorously Google goes about implementing this patent.