IoT news of the week for August 9, 2019

It’s the people, stupid: Behind any digital transformation are people who have to learn how to change the way they do their jobs thanks to new technology or new insights provided by technology. To get those people on board you have to build your IoT and AI projects with more than technical expertise. You need a good operator who understands what the people in the company will respond to. That’s the gist of this article, which goes into depth on the type of person successful companies need to have running their digital transformation. (Harvard Business Review)

Busting the idea that satellites can replace undersea cables: This is a good post that tries to debunk Elon Musk’s statement that satellites could handle more than half of the traffic that is now flowing through undersea cables. That’s a big claim, and it doesn’t actually pan out. (TeleGeography)

Wow, this is really terrible security: Leap, maker of a popular tablet for kids, recently had a series of vulnerabilities associated with their machine that are appalling. The company has patched the flaws, but the fact that data was sent from the device completely unencrypted feels like something a company should be fined for in this day and age. Additionally, the tablet had other vulnerabilities that could let people see the physical location of children playing with certain apps on the machine, and to start chats with them. I’m glad Leap did the right thing by patching the tablets, but I feel like more is needed here. This level of insecurity borders on negligence. (ZDNet)

A most excellent overview on surveillance: Anyone building an IoT product should read this and think hard about the type of data their device can collect and how they want to safeguard it. Not how they can share it with folks to make money, but how they can add value to the consumer of the device and protect that customer’s data. There is so much excitement over the possibility of using context clues to create an intuitive smart home or using devices in combination with AI to create personalized medicine, but in these examples, we’re essentially going to give our devices the ability to surveil us 24/7. Before that happens, we need to have a deep conversation about the harms of data gathering and aggregation. This article is a good place to start educating yourself. (Fast.ai)

SimpliSafe security system can be hacked with a $2 device: Well, this is unfortunate. It looks like a wireless signal emitter can “speak” the same frequency as the SimpliSafe security system and override signals from the sensors to the hub. SimpliSafe took issue with the video demonstrating the hack, saying it was misleading. A reporter at Cnet tried to hack its SimpliSafe System but was foiled. Cnet published a more detailed look noting that the attack is possible, but not plausible. (The Verge, Cnet)

It’s a hack attack! Defcon is this week in Vegas and I expect good security stories to subsequently come to light. I’m paying especially close attention to news from the Medical Device Village, where a model hospital room has been set up for testing and demonstration purposes. This article shares some of the efforts around making our medical devices and hospitals safer, something I’ve covered for years. I’m also eager to hear about the first-ever Aviation Hacking Village, which should lead to some scary stories but also signals a willingness on the part of aviation companies to take cybersecurity into the open (and hopefully make their efforts more robust). (Wired)

Are you building a platform or an ecosystem? Everyone wants to be a platform because platform companies can avoid commoditization if they are successful. Thanks to network effects, such companies can make it easy to get on board and build a business on a platform, then make it tough to leave. But this article points out that most platforms are more successful when they are actually ecosystems and share in the value created by all of the associated players. I wrote about this concept, but here are two researchers who studied Germany’s 370 platforms to lay out why some platform companies succeed and why some fail. (MIT Sloan)

A deep dive into where machine learning can fail and how to prevent it: A Google data scientist has written an article detailing common mistakes made when training machine learning models. It covers the commonly known case of a computer choosing to focus on a variable unrelated to the actual question the scientists are seeking an answer to, as well as how to split training and testing data. It’s dense, but essential for folks who want to understand machine learning and how it can go wrong. (Nature)

There’s a new LoRa modem with better software: Most radio tech develops in similar ways. Initially, chips are focused on the radios and how they will communicate with each other; most of the upper-level software is left to developers. As time passes and the software needs are further refined, some of the upper-level software gets standardized and placed onto modems. That’s where we are with LoRa. Semtech and Murata have released a new LoRa modem that handles some of the connection to applications and other software on the same device. (Semtech)

Walmart’s creating a smart home bridge: The device looks like it’s set up for the retailer’s InHome delivery service that will place groceries directly inside a customer’s home. (StaceyonIoT)

Where is Samsung’s smart speaker: Remember that fondue pot device that would pop Bixby, Samsung’s digital assistant, on a smart speaker shaped like a mini cauldron? It’s nowhere to be found. Samsung released two Note smartphones, an ARM-based Windows laptop, and a second-generation smartwatch this week, but it didn’t mention Bixby or the smart speaker.

Like this story? Get it and more every Friday when you sign up for my free IoT newsletter. Sign up here.