As more people create more online content; from status updates to reviews, mentioning brands and businesses; identity and trust are increasing in importance.
When faced with hundreds of comments about a restaurant, or thousands of reviews on a product you might buy, who the content was created by becomes much more important. Should you trust these people’s opinions? Currently, sites like Amazon and Yelp get around this by leveraging the “Wisdom of Crowds“. They aggregate reviews and present you with the number:
Amazon have evolved this format to present the “most helpful” reviews:
Yet, we don’t know much about the people doing the reviewing. In the Yelp example, what if the 749 people giving a 5 star review are not like me at all? And what if the 303 people giving a 3 star review are really like me? We can view the reviewers profile and previous reviews to determine a base level of trust, but the most important question is not whether they are trustworthy, it’s whether they value the things you value. 1000 aggregated reviews from a completely different demographic may be much less useful than 10 reviews from the same demographic as yourself.
The next evolution in online reviews will be the “Wisdom of Crowds Like Me”:
on Dec 9th, 2009 at 1:04 am
Pattern-matching combined with self-reported “likes” and ratings should go a long way.
TiVo supposedly can make excellent recommendations based on what I match, compared to what others are watching. Sometimes the key patterns are fairly stupid though (”We recommend all movies with Matt Damon!”).
Netflix also has this exact dual rating system in your sketch.
on Dec 9th, 2009 at 10:06 am
I think the key difference between now and sometime in the near future is that currently Netflix, TiVo and Amazon are only using one signal (Netflix and TiVo are looking at what you have watched, Amazon looking at what you have bought) whereas we’ll soon be aggregating lots of signals.
“People like me” will be aggregated from many dimensions of identity.
on Dec 9th, 2009 at 12:20 pm
Hiya Paul … I agree with your thoughts…
When reading movie reviews I find myself looking for a review by that reviewer of a film I’ve already seen .. giving me a benchmark on their tastes …Over time I’ll identify with the views and tastes of some but ignore those of others.
Perhaps Mark Hursts latest ‘Uncle Mark’ is a concentrated version of this.
http://goodexperience.com/2009/12/uncle-mark-2010-now-a.php
on Dec 9th, 2009 at 12:22 pm
Is the real problem that with all this information, with all this choice, we simply can’t make a fecking decision for ourselves (Adding Paradox of Choice to your Wisdom of Crowds)?
I can still remember a time where I relied on the one authoritative source for recommendations: Where will I go? Lonely Planet will tell me. Where will I eat? AA Gill will tell me. What will I drink? John Wilson (Irish Times) will tell me.
Sure, there’s all these “real voices” and people like me? But with all they’re contributing, maybe it’s harder, not easier to choose.
on Dec 9th, 2009 at 2:12 pm
You are onto something there Lar.
I’ve met people who have become so wedded to recommendation that almost every restaurant they eat in has been checked online for star ratings etc. beforehand …
Over time these people lose faith in their ability to make a decision, standing outside a restaurant googling for recommendations rather than using their own intuition.
Reliance on crowd wisdom = kissing goodbye to spontanaeity and serendipity ?
on Dec 10th, 2009 at 7:53 am
Thanks for the comments Nick and Lar!
Re. Paradox of Choice. This can be eliminated with good design. Choice exists all around us, yet we only suffer when presented with too many options. Just because the choices exist doesn’t mean that we need to present all of them in the UI. We need these systems because exponential choice isn’t going away. The ongoing technological development of networks will provide us with more and more information.
Re. One authoritative source.
There are two problems with “authoritative sources”:
- They can attest to quality of a product or service, but they can’t attest to personal taste. That’s why “people like me” is a powerful concept.
- It’s a dirty little secret in the travel industry that reviewers of services and writers of travel books often get paid to write good reviews. Even some of those that write for the large publications.
At the end of the day, whether we make choices based on instinct or technology will come down to the risk/effort and social norms.
Risk/effort: Choosing a restaurant by instinct is usually low risk, we may end up with a bad meal but it’s rarely very expensive, and hours later we’re eating again. But some meals are higher risk, like the family reunion. So it may be worth more effort to lower the risk. This scales all the way up to buying a car - choosing a car by instinct is high risk, higher financial commitment.
Social norms: It may become accepted that one would never choose something without checking it online first. Or a social norm might develop where people who do that for low risk decisions look like idiots to their friends.
on Dec 10th, 2009 at 2:37 pm
Paul: great insight and discussion.
If I may be so bold, the challenge you’re describing is exactly what we’re trying to solve at Hunch, a decision-making site whose content is crowdsourced.
Hunch’s decision results are based on an algorithm that looks at the preferences and tastes of ‘people like you’. So in proposing a car you might like, Hunch might ask you obvious car-related questions like whether you’ll be transporting kids or how much you care about guess mileage. But it also might ask you things like whether you have anything from Gucci in your closet or which type of politician you tend to support - questions that the algorithm has determined are correlated to people’s ultimate car preference. You can give it a try here: http://hunch.com/cars/
In fact, on every decision ‘result’ page, we show the popularity of the result among ‘everyone’ and also among ‘people like you’, exactly as your sketch above proposes. http://hunch.com/cars/land-rover-lr3/1375874/
Keep up the good work!
on Dec 17th, 2009 at 5:17 pm
re: ““People like me” will be aggregated from many dimensions of identity.”
This assumes that identity is singular. I think we each contain many aspects of identity (we don’t act the same way with our parents as we do with our college buddies) that may not make sense to aggregate.
Also, our social networks’ authority is not singular; I may trust one friend’s recommendation for stereos, but not for restaurants, and vice versa for another.
Maybe it’s just me, but I love the aspect of the Internet that lets me have multiple identities - a professional one on LinkedIn, a public-facing one in my blog, an informal one on LiveJournal, and an unmaintained one on Facebook :). I don’t want tools to aggregate my identity, because that would defeat the disaggregation I have performed on the Internet.
on Jan 10th, 2010 at 4:19 pm
Nice post Paul,
I couldn’t agree more with you, as UGC’s growing in quantity, it very often looses in perceived quality for the end user as there is too much data to process. And as a user, you have no simple mean to estimate how much you should trust someone’s review. If there was a way to assess taste similarities, maybe it would be a good start
At Qype.com (think the equivalent of Yelp in Europe if you don’t know them) , this is becoming a reality thanks to the use of the LikeCube innovative recommendation solution (disclaimer: I work for LikeCube).
In Qype, as a registered user with at least 3-4 reviews, you now get personalized recommendations of places based on your taste. You also get to see the people that have most influenced these recommendations and who’s reviews should be most relevant to you. These people are your taste neighboors for this recommended place.
Qype also offers a nice way to compare how similar you are with each person by looking at how you have rated places you have in common.
What we have done with Qype addresses the point raised by Eric, “I may trust one friend’s recommendation for stereos, but not for restaurants, and vice versa for another”.
In Qype, your closest taste neighbour isn’t absolute. He depends on the place being recommended.
There is still plenty of work to do around this, but given the initial feedback from qypers (some saying recommendations are “quite spooky”, “bang on the money” or “not seen anywhere else”), I believe what you call the “wisdom of the crowd like me” is part of the future of internet.