When did Amazon become so problematic?

Amazon this week kicked off a three-week conference dedicated to its suite of cloud computing offerings, collectively known as Amazon Web Services (AWS). Developers, analysts, and members of the media have signed up in order to understand what those offerings entail and what implementing them (there were roughly a dozen new ones launched during the initial keynote alone!) will mean for countless businesses that Amazon crushed by turning their products into AWS features.

While Amazon’s cloud offerings don’t generate much ire outside of those folks who are cranky about open source in general and the startups that suddenly find themselves competing against Amazon while paying the company to host their services, two of its existing offerings are getting lambasted against the backdrop of the AWS lovefest. The first is Amazon’s new Low-Power Wide-Area Network (LPWAN), called Sidewalk. The second is Amazon’s wearable device, known as Halo.

Both have become lightning rods of controversy due to an increasing tension between our desire for personalized, so-called “smart” devices and services and the data that companies need in order to deliver those services. Indeed, companies pushing digital assistants and smart services are learning that, rather than simply enabling a transaction, they are actually entering into a relationship with customers, which — like any relationship — has to be built on trust.

And right now, Amazon has a trust deficit.

Sidewalk has the buyers of certain Amazon devices upset because the company has opted them, without asking, into using their broadband connection to help transfer messages on its Sidewalk network. Users are not just unhappy with Amazon for essentially using their broadband data without permission, however; they are also worried about the privacy controls (or lack thereof) on Amazon’s network and the resulting potential for surveillance.

With Halo, Amazon has created a fitness wearable that collects a lot of personal data about a user’s activity, heart rate, and sleep. But using Halo’s Tone feature, it goes even further, recording snippets of that user’s conversations and assessing that conversation to determine the user’s tone of voice.

To be clear, the Tone feature will not be activated unless users opt in to it. All of the recordings are deleted from the phone every 24 hours, and none of those recordings ever make it to the cloud. But the very fact that an Amazon device has the ability to hear everything you say makes some people deeply uncomfortable, never mind the fact that it would be using that data to understand how you feel. The concern is that Amazon would do this in order to sell you something.

The device can also be used to take a series of photos of a user in their underwear or other similarly scanty clothing to determine their body fat percentage. The resulting images can be deleted (it took me two weeks to steel myself to take these photos, and I deleted them immediately afterward), but Amazon also offers users the option to store the images and any related data on their device or upload them to the cloud if they want to ensure their permanence.

As to why the company provides this option in the first place, Kara Swisher, tech’s doyenne, writes:

“In the last few weeks of using Halo, it finally clicked with me as to why Amazon needs a device that tracks sleep and movement and body fat and even body tone: An Echo is too far away from our bodies, and the consumer goods we order give the company much information about us but not enough.”

Her take is not that of someone in a healthy relationship with a connected product.

For a long time, the purported solution to lack of trust in a product and its data retention policies was transparency. Device makers believed that simply explaining to users where their data was stored and how it was being used would maintain their trust. To that end, Amazon is transparent about what Halo collects and where the data is either stored or dumped.

And with Sidewalk, Amazon has both published a white paper discussing the privacy and security of the network and made the general manager who heads it up available for deep-dives into what the company can and cannot see on it. Which, according to Manolo Arana, GM of Amazon’s Sidewalk Network, is nothing — Amazon can’t tell what data is being sent over the network or what devices are using it. And yet, at least one of my friends remains concerned that, thanks to Amazon’s network, the technology it runs on (a cheap LPWAN) could be used by the government to deploy trackers of its own.

To me, this is like getting angry at AT&T because a security agency uses its network to provide a GPS location from one of its trackers to the agency’s computers. Most people accept that AT&T is neutral in that particular instance, in part because they wouldn’t want to give up their ability to receive services like Waze or directions via Google Maps. Whereas people worried about the misuse of Amazon’s Sidewalk network are focused more on Amazon than the launch of what is essentially new underlying technical infrastructure in the form of a network for connected devices.

On the Halo front, for every person who finds the Tone feature creepy there is another person asking for more data. One of the features of Tone is that it marks certain moments throughout the day, letting a user know that, for example, for 1.6 seconds they said something that sounded like they were sad or discouraged. Upon learning of those moments, many people want to know more, specifically what made them sound sad or discouraged. I know I did. But when I asked Melissa Cha, a VP of product for the Halo device, about the ability to learn more, she said that to share that level of detail would mean storing the conversations, something Amazon isn’t doing.

Indeed, the pull between user privacy and deeper insights is one of the central tensions Amazon and other tech companies have to manage as we head toward a future with digital assistants trying to help us optimize our lives, our homes, and our health. We want our technology to know everything about us so those devices can provide better insights or new personalized services, but we also worry about how that data might be used against us or others.

Part of that tension stems from the fact that unlike a personal assistant, who knows most everything about us, a digital assistant doesn’t abide by any social contract. And the data it has can be used to benefit others, not just ourselves. Imagine how quickly an administrative assistant would get fired if they sold underlings the boss’s lunch schedule so those employees could “bump” into her and strike up a conversation.

Defusing the tension between the data we’re sharing as users and what companies will do with that data is essential to enabling wider adoption of connected devices and products. And to defuse that tension, we need to think about our purchases not as transactions, but as the start of relationships. The companies selling us those devices and services must do the same.

In any relationship, both sides have to trust one another. Traditionally when I purchased something from a manufacturer the relationship consisted of little more than a one-time interaction at a cash register, and any trust was tied primarily to any warranty expectations. I trusted that the product worked and the maker of the product trusted that I would use it in a reasonable manner. With connected devices, that manufacturer-user relationship has become much more complex.

And right now, when it comes to trust, Amazon is operating at a deficit with many of its users. The news that real people were listening to requests made to Amazon’s Echo devices shocked those who thought that only computers were involved. Meanwhile, for others, the casual approach to both security and surveillance embodied in the Ring devices and its Neighborhood app was the final straw. The news that Amazon was selling powerful surveillance tools to government agencies at a time when trust in those government agencies was at an all-time low also caused many customers to question their use of Amazon products.

I used to justify the amount of information Amazon had about me and accept new products from the company because I trusted its self-interest: Amazon wanted my data because it wanted to draw my attention to products online so it could sell me more stuff. I knew that Amazon was like a roach motel when it came to data (it went in, but it didn’t go out.) But I didn’t worry about Amazon contracting with third parties or my data ending up at a data broker.

As Amazon expands its businesses beyond selling me things and building a cloud computing business, however, I trust it far less. It feels like my relationship has become one in which I offer a lot of data in return for some value, but in the meantime Amazon is also using me to gather even more data and sharing that with people I would neither trust nor want to help.

For many people, their relationship with Amazon is fundamentally broken, and that affects their perception of all the company’s products. That said, I think Amazon is aware of some of these challenges. It’s clearly focusing more on transparency and is trying to explain how customers can control their data and avoid oversharing. But I also think certain relationships its businesses have with police departments and parts of the U.S. government taint its prospects, much like you might walk across the street to avoid a known sex offender or animal abuser.

There are two hard questions that need answering here: One, what should Amazon do to repair what has become a bad relationship? It could double down on selling the benefits of a smart home enabled by Alexa and safer neighborhoods enabled by Ring, while also trying to remake the health care system with a pharmacy service and a wearable device. And two, how should consumers react? Because it won’t convince everyone.

But maybe it doesn’t have to. For many consumers, the decision to get out of a bad relationship with a company providing connected services is tough. I shopped at local bookstores for Christmas presents this year, and found that only three out of the five books I wanted were available. Amazon had all five, and for the cost of four. I could dump Alexa or Google’s assistant and build my own local voice assistant, but it wouldn’t be as helpful or as accurate (plus I’d have to spend hours and money building and maintaining it).

And this challenge — the struggle to get out of a relationship with a badly behaving data-aggregating company that provides useful services that we rely on — is giving regulators fits. After all, when we talk about antitrust issues and data regulation, we are talking both about this tension between smarter service and privacy as well as the degradation of our relationships with the companies we once trusted.

The answer isn’t simply more transparency. It’s pushing companies to behave better, to insist that they use the data we grant them to improve our lives without also making us complicit in institutional racism or government-approved injustice. It’s also about clearing the way for companies to emerge with which consumers can create better relationships.

I am a big believer that data will unleash a plethora of new opportunities and benefits. But I need the companies to which I give my data to think that, too. And to act accordingly.