Gone Fishing

Gone Fishing is a research-prototype Android app that lets you and your partner know when you are not likely to read each other’s texts or messages.You can test the prototype by downloading Gone Fishing from Google Play.
Install its widget on your phone and you will see one on the following four visuals:

Screenshot AvailableBusiness as usual means that your partner is likely to pay attention to new messages. Note that the algorithm only predicts whether your partner is likely to see the message, but not whether s/he will write a response. Also be aware that the prediction is not 100% accurate. For these reasons, you may not get a response to a message even if this sign is shown.





Screenshot UnavailableGone Fishing means that your partner is likely to not pay attention to her/his phone and might not see messages that you send now any time soon.

Gone Fishing black & white means that your phone has not received any status updates from your partner’s phone in the last few minutes. This can for example happen if the partner’s phone does not have internet connection or if the Gone Fishing app crashed. You should not make any assumptions about your partner’s availability.



Screenshot Waiting

Waiting means that your phone is still waiting for the first update on your partner’s availability status. You should not make any assumptions about your partner’s availability.


First, download the Gone Fishing app from Google Play onto your Android phone: https://play.google.com/store/apps/details?id=org.pielot.cmm.

Screenshot SetupSecond, run the app to set it up.

Clicking the “Accessibility Settings” Button will take you to the Accessibility Settings menu. There, search for “Gone Fishing”, click it, and turn it ON. Press the back button until you are back to Call Me Maybe.

On some phones, you then have to click the “Notifications Settings” button, which will open the Notifications Access view. There, please make sure that the checkbox next to Gone Fishing is ticked. Again, press the back button until you are back to Gone Fishing.

Next, write down the 5-digit code (here ed502) and give it to your partner. Ask your partner for her/his code and enter it into the text field (here b5939). After that, you can close the app. You and your partner are connected now.

Third, add the Widget. To see your partner’s availability status, you need to install the Gone Fishing widget. Find the widget called Gone Fishing and place it into this empty field. The widget requires free space of the size of two app icons.

Once it is place, you should first see the Waiting image, which – if the setup on both phones worked – should change to “gone fishing” or “business as usual” after a while.

How the Prediction Works

Gone Fishing applies latest research results on predicting attentiveness to mobile phone messages and is still highly experimental. Basically, it feeds different values, such as the hour of the day or the status of your screen (on/off) into a machine-learning algorithm. We fed this algorithm with data from 48 users – alongside data on fast they actually attended to 30,000 messages. From this data, the algorithm derived rules on when people are typically (in)attentive. It relies on the fact that people typically exhibit distinct usage patterns when when they are attentive (e.g. having sent a lot of messages in the last minutes) vs not attentive to messages (e.g. phone is in silent mode and in the pocket). Of course, this kind of prediction is never 100% accurate, in particular when people have different than usual.


No personal information leaves your device. All computation regarding the attentiveness is done locally on your phone. With your partner, only the prediction (true/false) is shared.

To enable us to study how well the prediction performs, the app shares anonymous data on how well the prediction performed in the past. It logs whenever a notification from a messenger is received or attended to. When a notification is received, the phone logs the time when the notification arrived, the type of application, the status of the phone that is used for the prediction, and the results of the prediction. When the notification is attended to, it adds this to the log and sends it to a server. None of this data contains personal information, such as the content of the message.