Hacker News Reader

"It makes local hard to scale because your audience size is that much smaller that reviews are far less reliable both in frequency and volume"

I've been a member of Angie's List for over 2 years now. It's not only saved me countless amounts of money over choosing nationally-advertised service companies for stuff like plumbing and HVAC work, but it's saved me tons of aggravation filtering out the honest, hard-working businesses from the fly-by-night charlatans who pile on services and "annual maintenance contracts" that are just code for finding yet another problem next year.

How do they get "local" reviews? Simple. If you pay for it, you'll review to keep the cycle of reviews going. If you're invested $30-60/yr. to subscribe, you feel like you have a vested interest in keeping both the review site and the service provider alive. I liken it to the SomethingAwful forums: the ones who subscribe are also the most frequent posters.

But, Angie's List also uses their own tactics to keep reviews flowing and fresh. They will email reminders to review service providers you've recently browsed if you decided to use them. They'll even call you to record a review over the phone and then they'll transcribe it.

It's why they've been blowing everyone else out of the water in "local" search since the 90s.

I love the philosophy and have been keeping an eye on Angie's List for a long time but unfortunately it still isn't available in my area, although it looks like they are starting to think about Canada (province is mentioned here and there on their site).

What Yelp does is a racket, similar to what BBB does. In Yelp's case they are being helped by Google and other search engines. The problem is that when people link to them that's an upvote in Google's eyes. After enough upvotes are gained, provided they don't spam, there's little anybody can do to downvote them. It's not like you can negatively link to them. This, I feel, is one of the main problems with the like and +1 buttons. There's isn't a -1 or notlike option.

I suspect Yelp is extremely venerable and could be successfully sued from several angles. However, the court system tends to protect such companies on free speech grounds etc. So, their downfall will probably come from leaked internal operations rather than any external source.

A beautiful typo. You mean "vulnerable" for "venerable" right? Is that because you think they really are doing what the article implies? Is there evidence more substantial than that offered by the Tribune? The article for me is just shoddy journalism. Can it be that hard to really test the hypothesis that businesses that turn down the ad salesperson get screwed by Yelp? I don't want to defend Yelp since I don't like the site, but I think its pretty scummy of the Tribune also to willingly promulgate, at least by rhetorical implication, the idea that Justin G is some kind of shill for non-Brader dog trainers.

(Very much tangential at this point, so I'll put my theory that Justin G, far from being a shill, is actually a superhero in parentheses. Consider: He's very secretive. He reviews businesses without leaving behind a trace of his true identity. He doesn't respond to queries from the Tribune. And look at his profile: He wears a cape. A cape! Also, his reviews are annoyingly sincere. And as for his name, clearly Justin G is justing the world one review at a time. Yes, verbing weirds language, but if you can justly right the world, why not rightly just it too?)

http://www.yelp.com/user_details?userid=z25c9TyWY1TdVEGxkj34...

http://en.wikipedia.org/wiki/Conversion_(linguistics)#Humor

I've heard talk of dodgy behaviour from Yelp before. See eg. this article: http://www.eastbayexpress.com/ebx/yelp-and-the-business-of-e... (via here: http://www.jwz.org/blog/2009/02/yelp-shakedown/)

This isn't the first extortion attempt we've read about over the past few years, and Yelp's excuse is always their so called impartial "algorithm". Google, for their part, extorts Yelp in much the same way, as they used to do with Experts-exchange, until that racket became too obvious. If Yelp reviews were not so useless they'd probably have been sued for racketeering by now. That said I bet the first Federal RICO case they go up against will put them out of business.

As a consumer i'd prefer not getting any likes to getting tons of notlikes on my boring fb status updates...

Besides, an issue with such systems being used to determine merit is this... Which is worse, having a bunch of people who consume the content you created and not give a ditt or people connecting with it and disagreeing and marking a downvote

"the value of so-called "social search" is limited to nonexistent because most people in your circle won't review anything."

I completely agree. The other huge problem with social search is I don't have 100% overlap with all of my friends. If I like Alice's taste in music, and Bob's taste in food, searching among Alice's taste in food and Bob's taste in music is a complete loss for me. Instead, I want "Netflix search": "people similar to you that rated Foo 4 stars also rated Bar 4 stars.

Wait until someone links up all our friends' and coworkers' credit card purchases, netflix views, google searches, downloads, travel history, etc- or even just a few pieces of that- then you can start to come up with a recommendation engine (AKA social search) that doesn't require people to actively create content for.

I think social search has value if we define it differently. In other words, I agree with your assessment. The thing that is social, is search, and has value is something more implicit. It observes the behaviors and decisions of other people, perhaps your friends but perhaps not, and infers recommendations from those. Amazon and netflix already do this in their domains to great success. It will become more prevalent. This is not what most people think of when they say "social search," and this is not the thing that you are (rightly) deriding. But I would imagine that in a decade, this is what will eventually be described with the term "social search."

"The simpler you make it the more that will but the simpler it is the less useful it becomes"

Is there any data to support this point. Intuitively it feels that simple system would lead to higher number of reviews and consequently paint a more accurate picture of reality.

Intuitively it feels that simple system would lead to higher number of reviews and consequently paint a more accurate picture of reality.

It may lead to a higher number of snitty reviews, as in "I gave this restaurant 1 star because I went there for my friends birthday and she was depressed at getting oder." Or, "This book gets 1 star because UPS delivered it while I was not home and they left it in the rain."

"The problem is that comScore (and now Yelp) have an economic incentive to get people to advertise such that they are not impartial and site therefore cannot be trusted."

That's a key part of it. They do two things: keep track of a "fair score," and sell to some of the businesses being scored. It's an inherent conflict of interest. And they have not taken any steps to mitigate it.

Google faces the same problem, but the AdWords auction mechanism diffuses the COI so the temptation to extortion is mitigated.

As some have pointed out farther down, newspapers and magazines have the same problem. Serious newspapers have a wall between editorial and advertising to mitigate the COI.

>>It's why I think companies like Google (disclaimer: I work for Google) should stay out the content business.

Isn't AdSense content business?

Not really. And even if so, doesn't change the meaning or validity of the comment.

I question the value of social search in general. It doesn't matter what your friends like. What matters is what others with similar tastes like. The only good thing about social search is the context informed by the relationships.

Following you 100% on this vision.

There's been a lot of hype around social recommendation lately; it was fresh and new because only [somewhat] lately has 'Social' itself been made possible through Facebook.

But as I see things, we don't make friends because they have our tastes or because they can correctly recommend us stuff we'd like; we simply feel really good with friends, no matter what they like. Only the people who have our tastes can help us finding what we should like, or at least be able to do it better than most of our friends.

Basing recommendation solely on our social circle's tastes is like thinking our friends are the ones with the closest tastes to ours. And I think that's wrong.

Unless I got it wrong, Rex.ly for example, is taking a much better approach by letting you select specific friends to influence your recommendations.

> What matters is what others with similar tastes like.

Yes. Exactly. And it depends. Some friends have very different tastes than I - and sometimes that's exactly why they are my friends (who wants homogenous friends?).

For some items, I want to know what others with similar tastes like - like restaurants, for instance. A friend of mine loves fried food; me, not so much. But a stranger who likes healthy food may know of an unpopular restaurant none of my friends know of. That is exactly the kind of restaurant I would want surfaced in a good social search for restaurants.

For news articles, I would want the opposite. I don't necessarily want people with similar tastes to recommend articles to me. I'd rather read conflicting views across the gamut of the argument, so I can make my own (hopefully) informed decision. Again, friends aren't as good a source of such articles as strangers would be.

And to make it more complicated, that's just my preference. Someone else might prefer people with similar tastes instead.

I definitely agree that the context informed by the relationships is the key to social search - sometimes it's someone similar to you, other times, it's not.