To Our Users: A Community Update

//To Our Users: A Community Update

In the past couple months there have been a lot of changes at PressureNet as we’ve focused on becoming a profitable weather company. Today we want to talk to you about the journey that’s brought PressureNet to this point and share with you our vision for the future of this company, the app, and the community we’ve built together.

THE SHORT VERSION

If you don’t have time to go on a grand journey with us, the short version is that we have taken steps towards selling some of the archived and live barometric pressure data that we collect through the PressureNet app and our app partners. This is the first time we’ve ever actively sought out revenue in this way. We want to emphasize that protecting your privacy is important to us, and we invite you to contact us if you have any questions or concerns.

THE LONG VERSION

PressureNet began three and half years ago with a vision of building the largest ever network of barometric sensors. It’s phenomenal what can be accomplished using only software: you can sit at a computer anywhere and deploy weather sensor networks that have millions of nodes. PressureNet launched in late 2011 for the Motorola Xoom. It had just two features: 1) tap a button to send a reading, and 2) view other readings. We had – and still have – ambitions of eventually making the Best Weather App EverTM but people took to our idea even with the initial limited usability. We immediately found a passionate and dedicated group of users, many of whom are still with us today.

With the launch of the Samsung Galaxy Nexus later that year, barometers in smartphones became more widespread. Our number of users grew until we rivalled the density of existing weather sensor networks. Around this time we began working with our favourite scientist, Dr. Cliff Mass at the University of Washington. He was already researching the utility of assimilating dense pressure observations from smartphones into forecasting models, so we began sharing data and working to improve his results.

Such an undertaking requires solving a series of difficult problems, all of which stem from a simple core property of this data: it’s noisy. Traditional weather sensor networks have stationary observation tools that are built with high-quality and trusted components. They are held to a strict standard. On the other hand, smartphones move about constantly and the embedded barometers have a larger-than-usual error range.

To combat this noise and clean the data we would need to unite millions of users to help us cover the world with pressure sensors and receive a sufficient density of data.

We had a core of dedicated users as passionate and excited as we were about our vision, but the question remained: how do you get not thousands but millions of people excited about a barometer network app? (Hint: you don’t.) One solution is to expand it into the Best Weather App EverTM, but unfortunately that is a time-consuming task, made more difficult by the fact that there are already many popular weather apps. A quicker and less risky way of expanding a network to include millions of people is to connect existing networks of tens of thousands of people.

In 2014, we began to reach out to other weather app developers. We asked if they would like to join our network by integrating some of our code and contributing pressure data to PressureNet the same way our own app does. Two apps in particular were excited about the idea: Smart Thermometer by Color Tiger and Beautiful Widgets by LevelUp Studio. With the integration of these two apps our sensor network grew from 15,000 to around 400,000 devices!

We are now collecting five million measurements each day and have built up an archive of over one billion readings. This puts us in a position to make a crucial difference in weather forecasts.

MAKING THE BEST WEATHER FORECAST EVERTM

There are two main aspects to improving weather forecasts. The first is to take more measurements of the atmosphere (that’s us, and you!), and the second is to run better models on faster computers. Typically these activities are expensive: both weather observation satellites and supercomputers cost at least tens of millions of dollars.

PressureNet has changed the atmosphere observation industry by deploying a vast sensor network at very low cost. With the recent ubiquity of smartphones and the proliferation of sensors inside, we can observe the atmosphere at high-resolution without physically building and deploying expensive sensor networks. This means that we are in a unique position to run The Best Weather Forecast EverTM and have a revolutionary impact on the state of weather science.

Making The Best Weather Forecast EverTM is going to be difficult, so we’re taking it one step at a time. The first step is for our company to achieve financial stability. In the past our strategy has been to pursue investment, but so far we haven’t received an offer that has felt right for our vision of the company’s future. So we’ve decided to try another approach.

WHAT WE’RE DOING

What is the new approach? Recently PressureNet has been in discussions with our API users as well as other weather and environmental consulting companies. We’ve offered to sell our data as a commercial service for use in forecasting or research. We’ve received some interest. Many of you have been part of PressureNet from the very beginning (we love you!) and we would like to know what you think about us taking this direction.

Although this is a radical step for us, we have received great feedback through our social media channels, where we have been voicing our plans to commercialise our data we have been collected.  We think, and others tend to agree, that there is no point in simply stockpiling data if nothing is going to happen with it.  We decided that the best approach was to employ smart engagement software through Instagram bots to better reach our audience to understand exactly what it is that our users want.

By using these channels we have also generated substantial community interest, which is required to keep the data diverse and relevant.  Our goal is to have a global network of users, all sending us fresh data so that we can continue doing what we do.

WHAT WE’LL DO WITH THE MONEY

Earning revenue allows us to do a lot of exciting things (as well as some boring but necessary ones). Some of them are immediate; PressureNet has grown a lot and we can no longer keep up with server costs at our ever-increasing size. We can bring on more of our team as full-time employees. Other things we could do are more big picture; with the costs of expanding taken care of we could turn our full attention to cleaning and finding applications for our data.

A reliable high-density barometric pressure network has implications for many industries. For example, migraines have been linked with shifts in atmospheric pressure. In fact, some people use PressureNet to predict the onset of barometric pressure headaches. With a clean data feed, PressureNet’s data could be useful in medical studies on the link between air pressure and migraines, or simply be a more reliable tool for those who choose to use our app for migraine tracking.

The reason that barometers were initially embedded in smartphones was to help determine the phone’s altitude for GPS and geolocation purposes. The technology has not yet been perfected and a database of locations and pressure readings could be invaluable for this area of research. Location services are currently available in a variety of apps but only in a rudimentary form. For example, many fitness apps (such as for hiking or mountain climbing) are in dire need of more accurate altitudes.

Of course, the most direct application of quality pressure data would be for weather forecasting. While our raw, noisy data has already shown itself to be useful for forecasting by academic researchers, it has to go through slow quality control mechanisms before being fed into forecasting models. To bring about improvements in the real-time models run by industry forecasters our data would need to arrive to them already clean and quality-controlled.

WHAT DO OUR UPCOMING CHANGES MEAN FOR YOU AND YOUR PRIVACY?

If you use our data API, it will no longer be freely available once we start selling it as a commercial service. The PressureNet app, on the other hand, will remain free to download, free to use, and free of ads.

We are aware of the sensitive nature of selling user-contributed data and we want to be open about exactly what information we collect and what control you have over it.

The data is anonymous and is comprised of: an alphanumerical user id that is not directly linked to any personal user information, atmospheric pressure, location (latitude, longitude, and altitude) and time1 of the pressure reading, phone model type, whether the phone was charging at the time the reading was sent, as well as some other metadata2. PressureNet does not and has never collected any personally identifiable information.

As a user of our app you have a choice about who has access to the data collected from your device. If you do not want your data to be available outside of your phone, outside of PressureNet, or outside of research purposes, those options are available to you through the settings menu in the app. We also encourage you to read our updated privacy policy.

It goes without saying (but we’ll say it anyway) that PressureNet wouldn’t be where it is without the support of its users (that’s you). You are enormously important to us and we always want to hear from you. (Even though sometimes we’re bad at replying in a timely manner. Sorry!) If you have any questions, concerns, or comments for us we invite you to leave a comment below or write to us at software@cumulonimbus.ca and we’ll do our best to respond. We hope you are as excited about the future of this company as we are!

1. and time zone ↩

2. the PressureNet SDK version that the phone is running, whether the SDK is embedded in a partner app or in our own PressureNet app, the source of the location data (either GPS or network), and the phone’s confidence about the location fields (see the Android documentation for details about location confidence values). A list of fields returned by our data API is available in our documentation (in the “Response” section). You can also check out the PressureNet app and SDK source code on Github if you are so inclined.↩