Who can you trust?

XVI.2 March + April 2009
Page: 32
Digital Citation

Ps and QsOn trusting your socks to find each other

Elizabeth Churchill

back to top 

I was recently told that we are moving toward a world of "the Internet of Things." I affectionately call this "the arrival of the IoTs" (pronounced "eeyuts"). It seems this revolution will be most helpful specifically in the creation of the "aware home."

For example, if I am traveling, I can still vigilantly watch what is going on in my abode. I can control ventilation and heating. My front door will open for approved visitors even when I am not there. And hopefully it won't be long before all my misplaced possessions start messaging me about their whereabouts. Going further, we all know that ordering food over the Internet has become commonplace. However, imagine if my fridge decides I don't have enough food and sends me a message to ask if it should place a delivery food order. It could even place the order without consulting me, with delivery set for my usual arrival. Maybe my laundry basket will start crowing for attention because it is too full. Currently, these systems work separately, but we are fast approaching a world in which the systems will speak to one another. Goodness knows what will happen when they all start talking to each other. Am I going to be embarrassed when my fridge and laundry basket confer and decide I fit the programmed-for persona of a slob and therefore require an upgrade of commodity decision-making on my behalf? This is a whole different kind of commodity product agency.

Before I continue elaborating scenarios for this emerging IoTs world, it is worth saying a little more about what is meant by the "Internet of Things." A somewhat rough Wikipedia entry states the following:

"In computing, the Internet of Things refers to a, usually wireless and self-configuring, wireless network between objects, such as household appliances."

It goes on to speculate that such connected objects would be things like "cans, books, shoes or parts of cars," all of which would be "equipped with minuscule identifying devices."

How is this going to be possible? A proposed change is the move from IPv4 (Internet Protocol Version 4), the infrastructure of our current Internet world, to IPv6 (Internet Protocol Version 6). IPv4 was completed in the 1970s, and many networking experts believe we are almost out of the four billion addresses that are available in IPv4. IPv6 offers expanded addressing, moving from a 32-bit address to a 128-bit addressing method, so we can identify many more objects. Although the driving scenario for the enthusiastic Wikipedian authors is the reduction of stock shortages and wasted products, the dream has more layers.

The weak IoTs hypothesis (version lite) is that most objects will be addressable so devices can be "pinged" to see where they are and what they are up to. In the version-lite world, this is likely to be a network of dumb things that can be pinged and located; these locatable objects can't, except in the most minimal ways, answer back. Furthermore, aside from the most rudimentary data exchange, there will be little connection between the objects in the network. These objects will not be able to make decisions for themselves or chatter amongst themselves.

The strong IoTs hypothesis (the "full fat" version) includes the world of "spime"—a concept invoked first by science fiction writer Bruce Sterling. In 2004 Sterling painted an image of an interactive device that is enhanced with RFID and GPS tracking and can thus track its history of use. As more objects become addressable and develop more intelligence and agency, we will have a world of autonomous, sort-of sentient devices that communicate amongst themselves and will be able to auto-organize depending on the context. Some pundits of what has been called "ambient intelligence" are very excited about this version of the IoT world.

Nosey people still exist, but these days their options for snooping surreptitiously are so much greater. Curtain flickers not need even approach the window, so there are few cues as to who is monitoring your actions.

I am largely in agreement; this all sounds really exciting. My favorite, desired scenario for all this auto-organization calls for the development of sentient socks that can find each other. Yes indeed, I want a sock drawer that resembles Noah's ark, with neatly assembled socks stacked two by two. Right now what I have is a lot of singletons wondering where their other half went.

I have been spinning this kind of simple, everyday fantasy for a while. Years ago, Les Nelson, Tomas Sokoler, and I designed a suite of objects called "Tools That Tell Tales." One such tale-telling tool would be the loaned wheely bag that reports back to you to say it is having a nice time on vacation with your friend. Perhaps that wheely bag is a spime—but when we elaborated this design space of chattering tools, the term had not yet been coined.

One thing to note in our scenario, however, was that the tools told you their tales only when you asked for them. We never tackled how on earth they would know when and whether or not to share their experiences spontaneously with us humans or with each other, should the situation so demand.

I realize there are fundamental concerns about the autonomy, politeness, and social decision-making of these semi-sentient, communicating things. I am not really sure I trust my socks to self-organize without disrupting the other inhabitants of the clothing drawer. And what if my confused and lonely socks get so distraught in their unsatisfactory search that they get into a fight with each other and with my other objects and they collectively crash the operating system? As I think about whether I would or would not trust my semi-sentient socks, I realize that, for me, the cloud on the horizon of this dream world of sentient (or at least semi-sentient) objects is trust in all its forms.

Trust is a slippery concept. Judd Antin of the iSchool at UC Berkeley and I checked out the stats: The word has appeared in the titles of papers indexed by the ACM Digital Library more times between 2005 and 2007 (149 times) than in the previous seven years combined (1998–2006, 131 times). Research into trust is all about uncertainty and risk. Most of the reported research addresses trust in enterprises, especially in the context of e-commerce, trust as developed in mediated human-human communication contexts, or systems perspectives on trusted/untrusted networks and network security. In interface and interaction design, trust unpacks to the familiar concepts of reliability, predictability, credibility, and visibility/transparency.

I see at least three dimensions of uncertainty and risk for IoTs to address if they are to be deemed trustworthy by experiencers (these are not necessarily users, after all; we may just be experiencing these IoTs unknowingly—the word "use" implies awareness).

First there's system reliability, consistency, credibility, and transparency. As system designers, we know that people will not continue to use technologies that they cannot trust to do the job they are supposed to do on a regular and predictable basis. The problem is, once there has been a breach we could not have foreseen, distrust sets in. And distrust is much harder than trust to navigate. Distrust is about fear and self-protection; it is about not believing in the product, the tool. Once someone distrusts a system, it is very difficult to regain their confidence. Lack of reliability and consistency are deal breakers for most people.

Let's think about some design challenges that must be addressed to ensure continuing user trust in the home IoTs system: exception handlers for sock drawers; incompatible sock releases; house virus updates; and operating system conflicts for merging households. And then there are the open design questions: How are you going to debug the house if it decides to lock the bathroom door with Auntie Elsie inside? How do you negotiate with your household IT administrator if she is 13 years old and angry at you for grounding her? And how do these systems inter-operate: What is in an effective decision-making hierarchy? Who gets to have the last say? Think of the power struggle if my applications are in conflict—my jeans in conflict with my T-shirts about which require my attention first. Anyone who has been around children who are arguing can understand the power struggles that come about between somewhat independent, sentient, and opinionated agents. And the question of all questions: how many buttons are there on an IoTs household remote control?

Second, there is a stickier problem—the reliability, consistency, credibility, and transparency of the network information transport; that is, the possibility for data/information leakage. Internet connections are often insecure, spewing data out and allowing others to see our activities, intentionally or by accident. The boundaries between walls are permeable. What if my sock resolutely, but incorrectly, pairs with a sock next door? And what if my fridge starts putting my favorite foodstuffs on my neighbor's shopping list? Do I really want them to know that my sophisticated palate requires at least two jars of peanut butter a week? The question is: Are my sentient objects going to know whom to share content with and whom not to?

There may not be a malevolent entity using these data or snarfing your bandwidth, but even opportunistic information observers may enjoy this. When I was growing up in the U.K., there was a term for nosey neighbors whose personal joy is to research other people's personal lives: these curious individuals are called "curtain flickers", known as such because most of their observations took place by peering out of their windows. One of the advantages of curtain flicking for the observed is that you can see the movement of the fabric indicating that you are being watched. Nosey people still exist, but these days their options for snooping surreptitiously are so much greater. Curtain flickers not need even approach the window, so there are few cues as to who is monitoring your actions.

Third, there is the thornier problem of malicious attack through deliberate and intentional hacks. The boundary of bricks and mortar is easy to see; unseen entryways are more difficult. Marketers of household cleaning products have for years been warning us of unseen dangers like germs and small creatures that can enter our home; the germs of tomorrow in the IoTs world are going to be those in service of humans with malicious intent. Frankly, once you get malicious or self-interested humans in the loop, all is likely to go to hell. It is worse than the days of yore, when shills and confidence tricksters used classic motivators—ego, greed, avarice, lust, in fact, all seven deadly sins—to trap us into giving away information that in other circumstances we would not share. Worse than these kinds of social cons are unseen attackers who steal personal information like bank account details and social security numbers, without ever interacting directly with us. In these cases we may not know for some time about a breach.

The second and third aspects of trust here revolve around the permeable boundaries that the Internet creates, and in the home setting, that means a whole new angle to perimeter security. The perimeters of the home have shifted, requiring new forms of vigilance. Of course, this crossing has been happening for some time with TV shows downloaded to TiVos and so on, but with newly developing aggregated services for living environments, more people are crossing the residential gateway. These Internet-enabled agents who are hack-able and live in an integrated world of data flows—where my sensitive information resides—make me feel vulnerable.

If you think I am being overly conservative, picture this. It's a chilly evening, and as you head to bed and snuggle down, you feel safe in the knowledge that the next morning will bring a nice strong cup of coffee to break the day. But, in a hilariously titled online June 2008 entry, "All Your Coffee Are Belong To Us," Slashdot posted the following: "Craig Wright discovered that the Jura F90 Coffee maker, with its honest-to-God Jura Internet Connection Kit, can be taken over by a remote attacker, who can cause the coffee to be weaker or stronger; change the amount of water per cup; or cause the machine to require service. Best yet, the software allows a remote attacker to gain access to the Windows XP system it is running on at the level of the user. An Internet-enabled, remote-controlled coffee machine and XP backdoor—what more could a hacker ask for?"

Whether this coffee pot hack actually ended up causing people problems or not, I don't know; I could not find any follow-up stories. But the unwitting Internet-enabled home device as a Trojan horse is surely something we can all imagine. One response to this is that we need to educate users—or better still, let's just insist that users "be more careful." Not going to happen. It is a strange thing that although we see the Internet as a risky place, we do not take steps to protect ourselves. Study after study shows that people do not secure their wireless home networks. And you only need to spend 20 minutes on social networking sites to find out way too much about a person—information that could help you breach the confidentiality of their personal data. Finding out a whole heap of stuff about a person is really easy with just a little technical expertise; most famously, Sarah Palin, the running mate of defeated John McCain in the 2008 election had her email hacked—her password and security questions were easily guessed from information available on the Internet.

Trust is fostered through reliability, predictability, transparency, assurance, and insurance—and it is a moving target in the design of all evolving complex systems. Perhaps nowhere more so than in the upcoming IoTs world. This IoTs world is going to involve a lot more emotional engagement with data devices. We will be increasingly intimate with our semi-sentient devices; we will weave them into our lives and entrust information to them—probably more than we entrust our data to social networking sites, which we expect to safeguard our precious data. And we are more likely to feel emotionally distraught and betrayed when we discover breaches. The affective bonds that we develop with these interactive things will likely mean that models of vigilance, which assume dispassionate security practices and emotionally uninvolved risk assessments, will be even more challenged. These models get stymied by trusted "friends" with access to our information. Conversely, these models do not account for the betrayal we feel when our interactive helpers allow themselves to be hacked. We can of course treat our things like children, expecting their boundary setting capabilities to be at about the level of a 5 year old. But that does not seem like the right model.

In any case, I suspect that continuous negotiation with the IoTs and with the network providers who control them is going to make me tense or tired or both. All this makes me want to jump headlong into a research agenda centered on infrastructure policy and on network security, and to actively promote a view of the IoTs world in terms of socio-technical, emotional networks of trust, reliability, and confidentiality. Not simply a world of consumer devices, simple and innocent nodes in the networks within and between which digital information flows. Right now, when thinking about the strong IoTs hypothesis at least, I am inclined to agree with J.K. Rowling's character Arthur Weasley in Harry Potter and the Chamber of Secrets when he said, "Never trust anything that can think for itself if you can't see where it keeps its brain."

On a positive note, there are new employment opportunities here: interior home integrators, managed home Internet services, remote Internet locksmiths, thing-programming specialists, and thing therapists specializing in human and device family and couples counseling.

back to top  Author

Dr. Elizabeth Churchill is a principal research scientist at Yahoo! Research leading research in social media. Originally a psychologist by training, for the past 15 years she has studied and designed technologies for effective social connection. At Yahoo, her work focuses on how Internet applications and services are woven into everyday lives. Obsessed with memory and sentiment, Churchill researches how people manage their digital and physical archives. Churchill rates herself a packrat, her greatest joy is an attic stuffed with memorabilia.

back to top  Footnotes

DOI: http://doi.acm.org/10.1145/1487632.1487640

back to top 

©2009 ACM  1072-5220/09/0300  $5.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2009 ACM, Inc.

Post Comment

No Comments Found