Cellular Carriers, Information Cascades, and LoRa
I recently finished listening to the book Algorithms to Live By which takes algorithms from computer science, economics, and other fields and applies them to appropriate life circumstances. As a data scientist, this book was a great nerd out session for me filled with great insights. One of the concepts that you’ve encountered and maybe even heard of is called the information cascade. Though not always applicable, you can think of this as one way to explain the snowball effect. It is one of many ways to answer the question of why people, companies, nations, friends, and others do something that as individuals they would never do but as group end up doing, despite having different preferences.
Information Cascades
The basic idea behind an information cascade can be explained by what could happen when ordering pizza as a group of friends, and no one wants to feel like they are intruding on the others. This isn’t a perfect example, but good enough to get the idea across. One friend in the group turns to another and asks what kind of pizza they want. This friend, wanting cheese, but thinking the others want pepperoni goes ahead and says pepperoni. Then the second friend, hearing this, adjusts her stated preference, despite actually wanting a plain ol’ cheese pizza, and goes along with pepperoni. It turns out the third and fourth also want the cheese pizza but thinking the others want pepperoni, they also along with it. So now, the whole group has decided to order pepperoni despite actually wanting cheese the whole time.
What happened is the public information—the group discussion—overruled the private information—what each person’s true preferences were. And despite everyone in actuality wanting the cheese pizza, they end up ordering pepperoni. This is an information cascade.
Snowballs
I can’t help but think a similar thing may be happening in the IoT space. More specifically, I think the information cascade, among other factors, may be an explanation as to why some carriers have chosen to deploy LoRa technology. This, despite its performance not being much better than aloha as shown in multiple studies (see here and here) and certainly not able to scale to support tens of billions of devices. To start the cascade, one carrier anxious to get into the IoT decides to do a LoRa deployment. Orange could be this carrier, which has publicly declared also, that LoRa is only a stopgap, a temporary patch until their real tech of choice, NB-IoT, is ready for prime time. But that gets buried under other press, as saying that kind of thing is, well, bad press. To continue the information cascade others see Orange choosing LoRa, assume some sort of vetting has been done, and also decide to get involved with LoRa in some way. And soon, nobody has done their homework, but because it’s easy to point to prior announcements (the public information) they go for it. Despite the fact that LoRa is, if anything, a Local Area Network technology.
I’ll be posting more on this phenomenon, but couldn’t help share the nerding out with you.