Man Versus Machine

June 3, 2008 ~ Sebastien Provencher ~ 7 Comments

On Saturday, I realized that my old wireless router had died of old age. Not knowing what model to buy next (I had bought the first one because it was the only one compatible with my old iPAQ Music Center), I turned to my social graph for an answer. I used Twitter to ask my 375 “followers”: “just realized my wireless router at home died. Any advice as to purchase of a new one?”

In a matter of a few hours, I quickly received many valid answers both from Twitter and from Facebook where my “tweets” are broadcasted to my 612 “friends”:

A Montreal web entrepreneur told me to buy the Linksys WRT160N with a link to the product page
The partner at the VC firm who funded Praized Media said I should buy an Airport Extreme if I’m using a Mac
A former Ubisoft colleague told me to make sure I update the firmware before declaring my router dead
Another former Ubisoft colleague suggested a SonicWall router along with a link
A high school friend told me to buy Linksys and said I shouldn’t pay more than $80.00

Now, I could have easily queried Google for such a search. I could have looked for “how to buy a wireless router“, found relevant web sites like About.com or eHow, and identify important product criteria that way (and associated brand/models). But you know what? Research takes time. Pinging my social graph took me 1 minute and I got five valid answers in a very short time.

It got me thinking about how the social graph is structured, in terms of ease of access. It’s very easy to access friends & family. You usually have their e-mail address and phone numbers handy. It’s a bit harder to reach the people that have a shared interest with you (community members, neighbors, former colleagues, etc.) and it’s usually very difficult to directly ask experts for their opinion (have you tried pinging a movie critic lately?). What if you could easily reach all these people to ask them anything? And what if everyone had tools to make it easy to answer?

It also got me thinking about the whole Man vs. Machine debate. Who do you trust most for information/recommendations? Man (a real human being answering your query) or Machine (an algorithm that’s surfacing relevant information)? It’s Facebook/Twitter/Friendfeed/Mahalo vs. Google/Yahoo/MSN. Coming from the business directory industry where word-of-mouth is often considered the biggest “competitor” (with social media, it’s becoming the biggest opportunity!), I tend to find human recommendations more relevant and more interesting.

In an article about social media monetization yesterday, eMarketer says that word-of-mouth might be a key way to monetize social media as “62% of marketing professionals told TNS Media Intelligence and Cymfony that creating word-of-mouth or viral campaigns has great potential to impact their business.”

That Man/Machine debate is age-old as you can see from this quote from a 1968 Time magazine article: “With the Depression, the machines that had once seemed so heroic to the prosperous ’20s were suddenly transformed into villains. As production lines slowed to a crawl and millions were thrown out of work, surrealists depicted nightmarish phantom treadmills and airplanes that were trapped like dragonflies.”

As we get closer to the singularity (defined as “a hypothesised point in the future variously characterized by the technological creation of self-improving intelligence, unprecedentedly rapid technological progress, or some combination of the two.”), I think we’ll get into more debates around the value of Man versus Machine (or maybe I’ve been watching too much Battlestar Galactica).

Update: Danny Sullivan talks about “Search 4.0: putting humans back in search“.

Six Takeaways from Kelsey ILM 07

December 5, 2007 ~ Sebastien Provencher ~ 8 Comments

Last week, I was in Los Angeles for the latest Kelsey Conference (ILM 07). We heard presentations from many interesting speakers, most notably Jake Winebaum from RHD, Jay Herratti from Citysearch, Chamath Palihapitiya from Facebook, Stuart McKelvey from TMP, John Hanke from Google and the always interesting Jason Calacanis from Mahalo.

Once again, I had the opportunity to meet and discuss with many of my local search and directory industry peers, making this conference a must-attend if you’re in the local search industry. It took me the a few days to come up with takeaways from the conference, not because there weren’t any, but because they were embedded deeply in the zeitgeist of the whole conference and needed to be extracted. After a “disappointing” 2006 (as reported in this post from SES Chicago), I think we’re at a new inflexion point for the local search industry. It was almost as if every stakeholder in the room had realized that things were not as they had seemed to be and that they were being more realistic and pragmatic about online local search.

Without further ado, here are my takeaways from Kelsey ILM 07:

People are finally realizing that it is very difficult to “do” local. Both advertiser and user markets are very fragmented and local initiatives do not always scale. If you’re not “native” to the local search market, the learning curve is huge.
Clearly, the online local market has not been cracked yet. There is no clear winner yet and we’re still many years away from glory days.
Local is going to be huge online but the various stakeholders need to work together. Players have to identify where are their core strengths and weaknesses and partner to fill the gaps (either through aggregation of technologies, content or sales). M&A should be on everyone’s mind as well. Expect a very active 2008 on that front.
We heard the second reality check coming from a directory publisher in a couple of months. Time is running out and it’s now time to execute.
Verticalization is starting to happen. People are realizing that there are user & advertiser differences between yellow pages headings. We might finally see some real segmentation in the industry (headings-based pricing, vertical sites, specific ad products and content, etc.) .
Call-tracking/pay-per-call is now a strategic pillar of local. To solve the media fragmentation issue, this offers a unified business model to aggregate various products together and simplify the sales process.
Mobile is still the holy grail of local search, coming soon, but not in 2008. Maybe 2009.

How the Web is Becoming a Big Word of Mouth Machine

August 27, 2007 ~ Sebastien Provencher ~ 11 Comments

The day started with Robert Scoble discussing how “social graph-based search” (Mahalo, Techmeme, Facebook, etc.) is going to beat Google and other search engines.

Scott Karp summarizes Robert’s points:

Humans can judge what’s missing from an aggregation of information on a topic
The key to effective human filtering is leveraging a “fabric of trusted individuals”/”people who are trusted and credible”
By connecting these trusted people through a social network, you can leverage that resulting social graph to validate trust and create network effects

Then, Karl Martino added:

(…) there is a growing role for “Trusted Human Editors In Filtering The Web”. Our friends, our families, our communities. Not just machines and algorithms. My favorite and fellow bloggers, Slashdot, Salon, the home page of the NYTimes, Philly Future, Shelley Powers, Scott himself, my news reader subscriptions, are all trusted humans, or representations of trusted humans, filtering the Web for me. So it
still comes down to trust – What organizations do we trust? What systems do we trust? What communities do we trust? What people do we trust?

What it means: I believe the web is slowly transforming itself into a big word of mouth machine. Social will eventually be embedded directly in the fabric of the world wide web. Media companies have an advantage today as they are a trusted source but those that resist the “socialization” of the web will be left behind. In the directory business, there is a saying that word of mouth is the biggest competitor out there. I think it can become the biggest opportunity in local search.

Social Search Stronger than Google in South Korea

July 5, 2007 ~ Sebastien Provencher ~ 2 Comments

I’ve been reading many articles about social search in the press in the last few months. Jimmy Wales’ Wikia (and to a lesser extent Jason Calacanis’ Mahalo) has been getting a lot of buzz and I’m not sure I saw the big potential until I read this article in today’s New York Times. Naver.com isthe leading search engine in South Korea with 77% of all web searches (vs. 1.7% for Google) and it’s leveraging social search.

Highlights:

When NHN, an online gaming company, set up the search portal in 1999, the site looked like a grocery store where most of the shelves were empty. Like Google, Naver found there simply was not enough Korean text in cyberspace to make a Korean search engine a viable business. “So we began creating Korean-language text,” said Lee Kyung Ryul, an NHN spokesman. “At Google, users basically look for data that already exists on the Internet. In South Korea, if you want to be a search engine, you have to create your own database.” The strategy was right on the money. In this country, where more than 70 percent of a population of 48 million use the Internet, most of them with high-speed connections, people do not just want information when they log on; they want a sense of community and the kind of human interaction provided by Naver’s “Knowledge iN” real-time question-and-answer platform. (…)

Each day, on average, 16 million people visit Naver — the name comes from the English words neighbor and navigator — keying 110 million queries into its standard Google-like search function. But Naver users also post an average of 44,000 questions a day through Knowledge iN, the interactive Q.&A. database. These receive about 110,000 answers, ranging from one-sentence replies to academic essays complete with footnotes. The format, which Naver introduced in 2002, has become a must-have feature for Korean search portals. The portals maintain the questions and answers in proprietary databases not shared with other portals or with search engines like Google. When a visitor to a portal does a Web search, its search engine yields relevant items from its own Q.&A. database along with traditional search results from news sites and Web pages. Naver has so far accumulated a user-generated database of 70 million entries. (…)

Google, which started its search service in the Korean language in 2000, introduced an upgraded Korean-language service in May. The new version deviates from Google’s celebrated bare-bones style. In South Korea, people prefer portal sites that resemble department stores, filled with eye-catching animation and multiple features. “It’s obvious to me that Korea is a great laboratory of the digital age,” Eric E. Schmidt, the chairman of Google, said in Seoul at the introduction of the new search service.

What it means: I’m starting to think social search has a great future but I also think it’s difficult to start from scratch like Wikia and Mahalo. I also think there might be an amazing opportunity out there for directory publishers (and anyone operating a local search site with a good amount of traffic) to launch a social search application to complement their current database of content. Who will be the first large-scale local social search site?

Social Search Stronger than Google in South Korea

July 5, 2007 ~ Sebastien Provencher ~ 2 Comments

Highlights:

When NHN, an online gaming company, set up the search portal in 1999, the site looked like a grocery store where most of the shelves were empty. Like Google, Naver found there simply was not enough Korean text in cyberspace to make a Korean search engine a viable business. “So we began creating Korean-language text,” said Lee Kyung Ryul, an NHN spokesman. “At Google, users basically look for data that already exists on the Internet. In South Korea, if you want to be a search engine, you have to create your own database.” The strategy was right on the money. In this country, where more than 70 percent of a population of 48 million use the Internet, most of them with high-speed connections, people do not just want information when they log on; they want a sense of community and the kind of human interaction provided by Naver’s “Knowledge iN” real-time question-and-answer platform. (…)

Each day, on average, 16 million people visit Naver — the name comes from the English words neighbor and navigator — keying 110 million queries into its standard Google-like search function. But Naver users also post an average of 44,000 questions a day through Knowledge iN, the interactive Q.&A. database. These receive about 110,000 answers, ranging from one-sentence replies to academic essays complete with footnotes. The format, which Naver introduced in 2002, has become a must-have feature for Korean search portals. The portals maintain the questions and answers in proprietary databases not shared with other portals or with search engines like Google. When a visitor to a portal does a Web search, its search engine yields relevant items from its own Q.&A. database along with traditional search results from news sites and Web pages. Naver has so far accumulated a user-generated database of 70 million entries. (…)

Google, which started its search service in the Korean language in 2000, introduced an upgraded Korean-language service in May. The new version deviates from Google’s celebrated bare-bones style. In South Korea, people prefer portal sites that resemble department stores, filled with eye-catching animation and multiple features. “It’s obvious to me that Korea is a great laboratory of the digital age,” Eric E. Schmidt, the chairman of Google, said in Seoul at the introduction of the new search service.