One of our StyleHop advisors is very involved with a company that leverages crowdsourcing to create better/faster/more accurate algorithms.  It’s really fun to think about how StyleHop could leverage crowdsourcing to accelerate the iteration of our wisdom of crowd fashion forecasting model.  ~DR

Link to Original Article

September 21, 2009, 10:15 am
Netflix Awards $1 Million Prize and Starts a New Contest
By Steve Lohr

Update | 1:45 p.m. Adding details announced Monday about the extremely close finish to the contest.

Netflix, the movie rental company, has decided its million-dollar-prize competition was such a good investment that it is planning another one.

The company’s challenge, begun in October 2006, was both geeky and formidable: come up with a recommendation software that could do a better job accurately predicting the movies customers would like than Netflix’s in-house software, Cinematch. To qualify for the prize, entries had to be at least 10 percent better than Cinematch.

The winner, formally announced Monday morning, is a seven-person team of statisticians, machine-learning experts and computer engineers from the United States, Austria, Canada and Israel. The multinational team calls itself BellKor’s Pragmatic Chaos. The group — a merger of teams — was the longtime frontrunner in the contest, and in late June it finally surpassed the 10 percent barrier. Under the rules of the contest, that set off a 30-day period in which other teams could try to beat them.

That, in turn, prompted a wave of mergers among competing teams, who joined forces at the last minute to try to top the leader. In late July, Netflix declared the contest over and said two teams had passed the 10-percent threshold, BellKor and the Ensemble, a global alliance with some 30 members. Netflix publicly said the finish was too close to call. But Netflix officials at the time privately informed BellKor it had won. Though further review of the algorithms by expert judges was needed, it certainly seemed BellKor was the winner, as it turned out to be.

But the race was even closer than had been thought, as Netflix’s chief executive, Reed Hastings, explained for the first time at a press conference in New York on Monday. The BellKor team presented its final submission 20 minutes before the deadline, Mr. Hastings said. Then, just before time ran out, The Ensemble made its last entry. The two were a dead tie, mathematically. But under contest rules, when there is a tie, the first team past the post wins.

“That 20 minutes was worth $1 million,” Mr. Hastings said.

The Netflix contest has been widely followed because its lessons could extend well beyond improving movie picks. The researchers from around the world were grappling with a huge data set — 100 million movie ratings — and the challenges of large-scale predictive modeling, which can be applied across the fields of science, commerce and politics.

The way teams came together, especially late in the contest, and the improved results that were achieved suggest that this kind of Internet-enabled approach, known as crowdsourcing, can be applied to complex scientific and business challenges.

That certainly seemed to be a principal lesson for the winners. The blending of different statistical and machine-learning techniques “only works well if you combine models that approach the problem differently,” said Chris Volinsky, a scientist at AT&T Research and a leader of the Bellkor team. “That’s why collaboration has been so effective, because different people approach problems differently.”

Yet the sort of sophisticated teamwork deployed in the Netflix contest, it seems, is a tricky business. Over three years, thousands of teams from 186 countries made submissions. Yet only two could breach the 10-percent hurdle. “Having these big collaborations may be great for innovation, but it’s very, very difficult,” said Greg McAlpin, a software consultant and a leader of the Ensemble. “Out of thousands, you have only two that succeeded. The big lesson for me was that most of those collaborations don’t work.”

The data set for the first contest was 100 million movie ratings, with the personally identifying information stripped off. Contestants worked with the data to try to predict what movies particular customers would prefer, and then their predictions were compared with how the customers actually did rate those movies later, on a scale of one to five stars.

The new contest is going to present the contestants with demographic and behavioral data, and they will be asked to model individuals’ “taste profiles,” the company said. The data set of more than 100 million entries will include information about renters’ ages, gender, ZIP codes, genre ratings and previously chosen movies. Unlike the first challenge, the contest will have no specific accuracy target. Instead, $500,000 will be awarded to the team in the lead after six months, and $500,000 to the leader after 18 months.

The payoff for Netflix? “Accurately predicting the movies Netflix members will love is a key component of our service,” said Neil Hunt, chief product officer.

From Ad Age:

How a Notoriously Closed Culture is Adjusting to Social Media

Posted by Katie Hillier on 09.10.09 @ 10:03 AM

Buzz over New York’s 2009 Mercedes-Benz Fashion Week is growing and the world’s top designers are taking center stage, introducing their long anticipated collections. This is how the fashion industry has operated for nearly 100 years: Designers secretly produce the fashions that they feel are most relevant to their lines, and retailers decide what items are most relevant to their customer base. But the fashion industry is at the verge of a tipping point — one that could change this system forever.

Read More

NEW YORK, Sept. 10 /PRNewswire/ — First Growth Venture Network, a mentoring program for high potential seed and early stage start-up tech companies, today announced the selection of 15 seed and early stage start-up tech companies for its first class, or “Inaugural Vintage.”
Rest of Story

Our friends at Crowdcast outline the limits of prediction markets:  blog.crowdcast.com

StyleHop is growing!  Please let your friends know about some really cool post-college internship opportunities at StyleHop:

Blog Editor:  https://www.stylehop.com/internships/blogeditor/

Social Media Coordinator:  https://www.stylehop.com/internships/socialmediacoordinator/

National Intern Director:  https://www.stylehop.com/internships/interndirector/

For College Students:  https://www.stylehop.com/internships/

David
follow on twitter:  http://twitter.com/dmreinke

I don’t know Brad Feld but I love his Venture Capital blog, Feld Thoughts.  He understands investing in early stage companies is first and foremost a people business.

Here is his latest post on Startup Karma:

Great Entrepreneurs Believe in Karma

I met with an entrepreneur yesterday that I hadn’t seen in a few years.  I originally met with her about five years ago when she was starting her company.  She’d been a very successful executive at a large company and had decided to jump into the entrepreneurial game and re-invent herself.  Her business has grown nicely – and profitably – without having raised very much money.

We mostly just caught up on how things have been going (we email back and forth periodically, but it had been a while since we had met in person.)  After about ten minutes, she asked if she could tell me a story about the first time we met.  Always game for a good story, I said sure.  It goes something like this.

I was introduced to you by someone I had met with who was a close friend of yours.  He suggested that I get together with you and made an introduction.  After I set up the meeting, I went online to learn more about you.  After poking around for a while, I suddenly got scared – I had no idea why I was going to meet with you or why you would bother meeting with me.  I didn’t want to blow my one meeting and waste your time.  I told this to the person that introduced us and he said “don’t worry about it – Brad will give you 20 minutes of his undivided attention and something good will come out of it.”  So I went ahead and met with you.

I was really nervous.  I was uncertain what to talk about and just starting telling you about my business idea.  You gave me some reactions and a few pieces of advice and as some point said “I bet you wonder why I am meeting with you.”  I had no clue, said so, and you responded, “I believe in karma.  When I was starting out as a first time entrepreneur a bunch of experienced people helped me, gave me advice, and just spent time with me with no particular expectations on their part, except to be helpful.  I’ve never forgotten that and want to pass it on.  I have no idea what will come of this conversation, but if I’m helpful to you, you can pay me back by being helpful to another first time entrepreneur after your become successful.” This has stuck with me from the very beginning of my business and I repeat it often.

This story made me smile a huge smile.  I remember all of the entrepreneurs that helped me early in my career, including guys like my dad, Gene Scott, Helena and Chris Aves, Stewart Forbes, and many others.  Whenever I help someone like the entrepreneur above, I’m paying others back for helping me.

Many of the great entrepreneurs I’ve met believe in this and practice it.  It’s not altruistic nor is it selfless as there are huge emotional returns from watching other people – who you’ve helped early in their entrepreneurial career – be successful.  If you are a multi-time entrepreneur, I encourage you to consider a daily (or weekly, or monthly – whatever works for you) karma break to help someone that is just getting started.

“Despite being around for 10 years, the personalized product recommendation market is still in its nascency,” says Mr. Grau.

Jeffrey Grau is the retail e-commerce analyst at eMarketer.  Jeff and I have talked about how social shopping is going to help consumers understand local fashion trends.  While social shopping has pretty much been a disaster so far, I think he’s absolutely right that we are just getting started.  Check out the full article here.

Since starting StyleHop I have spent a lot of time thinking about online games and we have been fortunate to have a couple of really talented gamers helping us develop fashion games women will love.  We have a couple of fashion games out now (Fashion Cents and Hot Tops Boutique) that are good but our next game, FriendTrend, scheduled to launch in April is going to be great.

So what makes a great online game?

1. Delight – That intangible, magical something you feel when you first encounter a game.
2. Originality – There has to be something unique, right?  But it can’t be too different.  It needs to feel familiar yet fresh.  A tough balancing act for game producers.
3. Hook – There’s a rhythm created that keeps you playing….over and over and over again.
4. Click, Click – We are learning this one….the more you get that mouse moving, the more you keep the brain engaged.

Let’s crowd-source the last six.  What do you think makes a great online game?

Richard MacManus’ post 5 Problems of Recommender Systems.

In a Guide to Recommender Systems, a blog post written by ReadWriteWeb founder and editor, Richard MacManus, 4 types of recommender systems are identified:

Personalized recommendation – recommend things based on the individual’s past behavior

Social recommendation – recommend things based on the past behavior of similar users

Item recommendation – recommend things based on the item itself

A combination of the three approaches above

Each of these approaches has distinct advantages and disadvantages and the post outlines how both Google and Amazon use a combination of all three to drive their recommendation engines.

As I look at the current state of online fashion it’s pretty clear no one has figured out how to leverage these recommender systems in fashion affiliate ecommerce.  There are simply too many styles, they have too short of product lives and the diversity of consumer wants and needs are too segmented.  The complexity in making fashion recommendations makes creating an algorthimic solution extremely difficult.  Past behavior is not a good tool because the trends are always changing.  Item recommendations don’t work because there are simply too many product attributes in fashion and each attribute (think fit, price, color, style, fabric, brand, etc) has a different level of importance at different times for the same consumer.  Social recommendations, though challenging, hold the most promise.  So far, though, social shopping has fallen flat.

The problem, I think, is trust and scale.  In fashion, unlike say electronics or travel, you aren’t going to trust just anyone’s advice.  You want to know who is making the fashion recommendations and they need to have some credibility.  To overcome this trust issue, many fashion recommendation sites focus on “friend recommendations”.  The problem here is this model doesn’t scale.

By example, consider TripAdvisor, if you are looking for resorts in Bermuda, you will see the top-ranked 4-star resorts at the top of your search. The top-ranked resorts have credibility as “being the best” in part because of the many, many lower ranked resorts further down the list. The consumer knows that all resorts have been carefully considered and these are the best. Friend recommendations don’t pass this scale test. You just don’t have confidence that all the styles out there have been discovered and considered. Therefore, the recommendation has much less weight and credibility.

At StyleHop we believe the way to get around the trust-scale tradeoff in fashion social recommendations is to let users explicitly identify their “fashion peer group” when they are performing product searches.  So in addition to the typical product attribute refinements you can make when shopping for clothes online, we want to allow you to filter your results based on the 5-star rankings (think netflix) of folks you trust to help you shop.  For a nineteen year old going to college in the fall for her freshman year, she may want to see the styles highly ranked by women at the University of Michigan.  For another woman moving to New York City for her first job after graduation, she may want to see the top ranked wear-to-work clothes of 20-something Manhattan women.

By allowing consumers to identify their fashion peers and adjust who their peers are for each search, the user is in control and the results are broad and deep. Most importantly, you are giving the shopper validation -before she makes the purchase- that what she is buying will be acceptable among her peer group.  This goes well beyond surfacing top styles the consumer might like and acts as a powerful social reinforcement of the purchase decision.