> Say you’re looking for a vacation destination. You can of course search the web, but what if you want to learn from the experiences your friends have had on their vacations? Just as in real life, your friends’ experiences are often so much more meaningful to you than impersonal content on the web.
No, they are not. What's meaningful to me are the experiences of people (anyone) who have been to where I'm thinking of going, and their average rating.
Imagine this on Amazon: when considering buying a book, would you want to see the reviews of "your friends", of which maybe one or two have read the book you're looking at and taken the time to comment on it, or would you rather have access to the collective knowledge of millions of Amazon customers...?
I disagree. Because the things I value are different from the things everyone else values in a vacation. Same reason I can look at the bestseller list in Amazon and be completely uninterested in most of the titles.
That's not to say that I have the same taste as all my friends, either. But it's true that their experiences are often much more meaningful, if only because I know them well and understand how they would come to the conclusions they came to regarding a specific review, and I can extract a lot of information from that.
More than that, though, I want Google to be able to learn from my preferences and my friends' preferences in order to point me to the most relevant content.
The collective knowledge of the internet as an aggregate, averaged out, is not useful to me. It doesn't become useful until it's filtered to suit my needs. Which has always been Google's value proposition.
Some of the recommendation engines out there (and it seems everyone has one these days) are really good at figuring out my tastes and preferences. Netflix, Amazon, and hunch.com (now owned by eBay) come to mind. hunch.com recommendations are freaky-good because they ask questions whose core intent is to learn your tastes.
You're right that my friends' tastes are only weakly correlated with my tastes. Assuming my friends and I have the same tastes is an error.
I'd also like to add that the reviews of experts are something I still value over the reviews of others. For example, I'll pay close attention to what Andy Ihnatko has to say about new Apple stuff. Rotten Tomatoes does something interesting in this regard: they keep the expert reviews (e.g. Roger Ebert) separated from the "everyone else" reviews.
If you think about it, a core idea of pagerank was to figure out which sites are most authoritative. Now the question becomes: who is the most authoritative on topic X?
Now the question becomes: who is the most authoritative on topic X?
I think that was Google's project. I think now it's: who is the most authoritative on topic X in a way that's personally relevant to me, even if I have never heard of that person before?
I sort of agree with you on professional reviewers, but I sort of don't. The dilemma of a professional reviewer is that as soon as you're a professional reviewer, you're not really a normal consumer anymore and your experience of a product is going to be very different. (For example, a common complaint about the Android eco-system from reviewers is that it's annoying for each manufacturer's user interface to be slightly different. But a normal consumer who buys a phone and uses it for the length of a contract really doesn't care about this so much as the quality of the skin itself.)
So, basically: I think professional and amateur reviews both have value.
More on topic, though: while I don't really subscribe to the "filter bubble" theory (I still think the digital medium exposes people to way more different perspectives and arguments than pre-digital), I do kind of wonder about the turtles-all-the-way-down aspect of all of this algorithmic ranking. I think it's a subtly different problem.
"Now the question becomes: who is the most authoritative on topic X?"
That question is also answered as part of this launch, as the "People and pages" bullet point. If you're in the fraction of users who this has rolled out to so far, try searching for [jquery]:
Imagine this on Amazon: when considering buying a book, would you want to see the reviews of "your friends", of which maybe one or two have read the book you're looking at and taken the time to comment on it, or would you rather have access to the collective knowledge of millions of Amazon customers...?
Ideally you would have both. There is no reason to choose only one.
In the Amazon case, it might make sense to put your friends reviews atop all the others, so that you can see them in a sea of hundreds of reviews.
But with search engine results, I'm not so sure if personal should automatically mean placement at the top - since so often we've conditioned ourselves to never need to look a few places beyond the top link.
I have to agree with that statement. Moreover this increasing please-dont-confront-me-with-views-outside-of-my-peer-group is problematic. While we embrace tolerance and open-mindedness in society, services like Facebook and now Google Search are increasingly reducing the open character and often randomness of the web by creating separated friend-islands.
I think this is something we naturally do as humans. I've been thinking about "small world" connections lately. People that I know through disparate connections, sometimes via very different parts of the country. Then I think about the similarities... we're all probably in the same marketing demographics, age, wealth, education, ideology, etc... We've self selected to increase our probability of meeting.
It's a small world because we self filter it that way. I wouldn't limit this to Facebook and Google, I'd add any taste-maker or collection sites like Pinterist or Tumblr as well.
I agree that we humans filter ourselves into a bubble, but the recommendation systems are designed to discourage you from doing different. It is in their best interests for you to be as predictable as possible. Getting their recommendation algorithms to be supremely intelligent and nailing your interests is the hard way, meeting the user in the middle by reducing the set and complexity of his interests is the other easier way.
I'm not railing against recommendation engines. They just rely on fundamental assumptions about the nature of our interests that I'm not sure are right. I'd be interested to see a recommendation engine that had an explicitly different goal in mind than predicting your interests (e.g. an in-built bias to encourage you to eat healthier when giving you restaurant recommendations).
Now that's an interesting idea and it seems like a logical next step - first to predict the behavior and then trying to implement a latter of behavior changing steps to ultimately modify it.
But on the other hand I have to admit that the current status seems still highly rudimentary. The service most likely with the most data on me is Amazon and despite an almost 10 year customer history their suggestions are still nowhere near what I am looking for.
It depends; my friends may not have so many reviews, but on the other hand, I know they aren't shills or idiots. Preferably, I want both, which is what Google seems to offer from now on.
its not an either/or situation. when I'm looking for reviews of a book, or travel recommendation, I'd like to see my friends reviews as well as the top rated public reviews. content from people I know gives me a reference point on how to judge their opinions.
If I find an interesting blog post about a vacation on Google+, I can add that author in my "vacation reviews" circle. Further vacation posts by that blogger would also show up on my feed.
Why wouldn't that be a good use of personalized results?
I'd like to see both. My friends might not have as comprehensive coverage, and they may not have my exact taste, but _I know who they are_. This allows me to make judgements about their opinions that I can't make with strangers.
It's also generally the case that people will favor the opinion of a few people they know over the opinion of hundreds of strangers. For example, you'll try a restaurant because one friend recommended it, but the threshold for where you'll follow positive Yelp reviews to go to a new place is probably much higher (in the hundreds for me.)
No, they are not. What's meaningful to me are the experiences of people (anyone) who have been to where I'm thinking of going, and their average rating.
Imagine this on Amazon: when considering buying a book, would you want to see the reviews of "your friends", of which maybe one or two have read the book you're looking at and taken the time to comment on it, or would you rather have access to the collective knowledge of millions of Amazon customers...?