Trust is Not Reputation
I trust my dog with my life – but not with my ham sandwich.
That is but one of dozens of humorous ways to indicate the multiple meanings we attach to the word “trust.” It’s remarkable how good we are at understanding the word in context, given its definitional complexity.
One interesting aspect of trust is its relationship to the concept of reputation. This issue is coming to the fore in the so-called “sharing economy” or “collaborative consumption” movement.
Who can you trust on the Internet to deliver the goods they said they would deliver (think eBay), to leave your apartment in good shape if you lease it on Airbnb, to not be a creep if you offer someone ride-sharing?
It’s tempting to look at the concept of reputation as the scalable, digital badge of trust that we might append to all kinds of transactions between strangers, rendering them all as trustworthy as your cousin. (Well, most cousins.)
Tempting, but not exactly right. Because trust, it turns out, is not reputation.
Greenspan’s Folly
William K. Black has written about the dire consequences of Alan Greenspan confusing trust and reputation, saying:
Alan Greenspan touted ‘reputation’ as the characteristic that made possible trust and free markets. He was dead wrong.
Greenspan believed that Wall Streeters’ regard for their own reputation meant that markets were the best guarantor of trust – because they would perceive their own self-interest as aligned with being perceived as trustworthy.
Unfortunately, Greenspan’s belief was probably based more in ideology than in history or psychology, as the passion for reputation was overwhelmed by the passion for filthy lucre, immortalized in the acronym IBGYBG (“I’ll be gone, you’ll be gone – let’s do the deal”).
Early Social Reputation Metrics
Think back, way back, to November, 2006. A company called RapLeaf was on to something. Here’s how they described their goal:
Rapleaf is a portable ratings system for commerce. Buyers, sellers and swappers can rate one another—thereby encouraging more trust and honesty. We hope Rapleaf can make it more profitable to be ethical.
You can immediately see the appeal of a reputation-based trust rating system. And with a nano-second more of thought, you can see how such a system could be easily abused. (“Hey, Joey – let’s get on this thing, you stuff the ballot box for me, I stuff it for you, bada-boom.”)
Then there’s Edelman PR’s pioneering product, TweetLevel. It does one smart thing, which is to avoid a single definition of whatever-you-wanna-call it. Instead, it breaks your single TweetLevel score into four components: influence, popularity, engagement, and trust.
Edelman says:
having a high trust score is considered by many to be more important than any other category. Trust can be measured by the number of times someone is happy to associate what you have said through them – in other words how often you are re-tweeted.
According to TweetLevel, here are my scores:
- Influence 73.4
- Popularity 70.1
- Engagement 56.4
- Trust 46.9
So much for my trustworthiness.
Guess who owns the number one trust score on TweetLevel: it’s Justin Bieber. Now you know who to call for – well, for something.
The KLOUT Effect
It’s easy to poke fun at metrics like TweetLevel that purport to measure trust; but in fairness, because trust is such a complex phenomenon, there really can be no one definition. What TweetLevel measures is indeed something – it’s not a random collection of data – and they have as much right to call it ‘trust’ as anyone else does. Indeed, I respect their decision to stay vague about what to call the composite metric.
KLOUT raises a more specific question: it directly claims to measure Influence, and is clear about its definition, at least at a high level:
The Klout Score measures influence [on a scale of 1 to 100] based on your ability to drive action. Every time you create content or engage you influence others. The Klout Score uses data from social networks in order to measure:
- True Reach: How many people you influence
- Amplification: How much you influence them
- Network Impact: The influence of your network
I find that to be a coherent definition. If I’m a consumer marketer, I want to know who has high KLOUT scores in certain areas, because if they drive action, I want them driving my action.
Note that Klout doesn’t mention reputation at all – just influence. Where does trust come in? Klout says, “Your customers don’t trust advertising, they trust their peers and influencers.”
Well, I wouldn’t go there. On TweetLevel, the top three influencers are Justin Bieber, Wyclef Jean, and Bella Thorne. Influencers – definitely. People to be trusted? What does that even mean?
Trust Metrics
One problem with linking trust to reputation is that it can be gamed. One problem with linking trust to influence is that notoriety and fame are cross-implicated. Bonny and Clyde were notorious, so was Bernie Madoff and the Notorious B.I.G. – that doesn’t make them trusted.
Take Kim Kardashian. Is she influential? You betcha: her Klout score is a whopping 92. Does she have a reputation? I bet her name recognition is higher than the President’s.
But – do you trust Kim Kardashian? Well, to do what? (By the way, TweetLevel gives her a 70.1 trust score – way higher than mine. Now you know who to ask when you need a trustworthy answer; I’m referring all queries to her).
So here are a few headlines on trust metrics.
- They’re contextual. You can’t say you trust someone without saying what you trust them for. I trust an eBay seller to sell me books, but I’m not going to trust him with my daughter’s phone number.
- They’re multi-layered. Both Klout and TweetLevel correctly recognize that social metrics can’t be monotonic – a single headline number is useful, but it had better have nuances and deconstructive capability.
- Behavior trumps reputation. You can get lots of people to stuff the ballot boxes for you; it’s a lot harder to fake your own behavioral history. Trust metrics based more on what you did, rather than just on what people say about you, are more solid.
- Good definitions are key. When people say ‘trust’ and don’t distinguish between trusting and being trusted, they’re not being clear. There’s social trust, transactional trust – it goes on and on. Good metrics start by being very clear.
So what’s the link between reputation, influence, and trust? There is no final arbiter of that question. Language is an evolving anthropological thing, and as Humpty Dumpty said, words mean what we choose to say they mean. So job one is to be clear about our intended meanings.
—-
Full disclosure: I have a small interest in a sharing economy company, TrustCloud. I have written more about the sharing economy and collaborative consumption in a White Paper: Trust and the Sharing Economy, a New Business Model.
What you are largely describing here is the ways we wish technology to take effort and
ersonal responsibility away from us. I.e. – wouldn’t it be great if the computer tells me who to trust, who to hire, who to buy from. Would’t it be great to not have to worry about those judgements because the computer takes the responsibility away from me personally, and invests it in some sort of cyber- process.
Trust – who we trust and who trusts us – drills down to the very core of what it is like to be human. Trust takes judgement, gut feelings, emotion, body language, and even love. All processes a computer application can’t compute as a mathmatical algorhythm. Thats why these web communities come up with such strange results. Take the human out of the process and judgement of who to trust becaome fickle. Look at how we made credit judgements that brought about the fiscal problems. When bankers were lending their own real reality money, money from their own personal accounts, bankers didn’t lend to people who when talking to them face to face, they didn’t trust. Take the face to face component from the process of building trust amd the results are unpredictable.
Building and creating trusted networks is time consuming because of the face to face element. Maths algorhythms will never replace us in this role. Therefore a web or computer system that tries to do the job will generally have less personal judgement than my dog. He loves anyone who has a biscuit – even baddies.
Wish I could find the way to correct my typos?!