The Noisy Channel

 

Looking for a Devil’s Advocate

March 26th, 2009 · 62 Comments · Uncategorized

I blog a lot about the virtues of exploratory search and the narrow-mindedness that Google and others exhibit in their focus on ranked search result lists, as well as about my skepticism about the value and longevity of the ad-supported model. I think it’s safe to say that I’m largely preaching to the converted–if anything the comments often amplify rather than challenge my premises, or even go further, arguing that I’m understating my arguments.

On one hand, it’s reassuring to hear validation for deeply held but hardly uncontroversial views. On the other hand, I’d love to find passionate advocates for the other sides of these issues who are interested in debating them. I’d like to believe I’m open-minded, even about the beliefs I hold most deeply, and in any case that I’ll offer an opposing argument a fair hearing and serious consideration. I trust that readers here would be just as fair-minded.

The problem is that those advocates don’t seem to show up, even in the comment threads. Perhaps that is because of my incredible powers of persuasion, but I imagine that no one wants to be a lightening rod for criticism, so those with opposing views lurk quietly or simply take their eyeballs elsewhere.

What to do? Is there any chance that someone who strongly disagrees with me on at least one the aforementioned points would be interested in writing a guest post? I’d even be willing to post it anonymously. I just want to make sure we’re not getting into the intellectual rut of a mutual admiration society.

Please contact me if you are interested; this is a serious offer, and I’m quite open to suggestions on how to make it work.

62 responses so far ↓

  • 1 Christopher // Mar 26, 2009 at 12:33 am

    “I just want to make sure we’re not getting into the intellectual rut of a mutual admiration society.”

    Ok, great line! 🙂

  • 2 Daniel Tunkelang // Mar 26, 2009 at 12:37 am

    I came in here for an argument!

  • 3 Christopher // Mar 26, 2009 at 12:40 am

    Why I thought my comment was particularly humorous. 😉

  • 4 Morton Swimmer // Mar 26, 2009 at 12:41 am

    I’ll bite. First of all, I don’t think that Exploratory Search and Ad-supported capitalization are mutually exclusive. In fact, targeted ads, while scary, can be useful and might be a legitimate conclusion of an Exploratory Search. What we all hate is the spam, so anything that increases the relevance of what we see is a Good Thing ™.
    So, while I am totally with you on the subject of Exploratory Search, I don’t see why an Ad-supported capitalization model doesn’t have a future.
    Cheers, Morton

  • 5 jeremy // Mar 26, 2009 at 12:42 am

    On one hand, it’s reassuring to hear validation for deeply held but hardly uncontroversial views.

    Maybe because deep down, even your opponents know that you’re right. They know that they really should be doing exploratory search, but don’t, because it would take away from their one and only revenue stream: advertising.

    So maybe there is no passion and conviction on the other side.. just secret guilt, wrapped up inside of an ever-weakening facade. I mean, how long can anyone continue to say “yes, we put ads on top of the organic results, because that serves the users better”, with a straight face?

    I dunno. That’s just a thought.

    I agree thought; it would be good to get dissenting voices.

    Wait, did I just amplify your premise, yet again? Apologies 😉

  • 6 jeremy // Mar 26, 2009 at 12:46 am

    Morton, what do you mean, exactly, by ads being a legitimate conclusion of an exploratory search? Do you mean that, after having a robust and thorough interaction with all the possible facets and avenues and consequences of one’s information need, the advertisement might be the item that is most relevant? Is that what you’re saying?

  • 7 Daniel Tunkelang // Mar 26, 2009 at 12:54 am

    Jeremy, it’s ok if you make me feel like a wimpy moderate. 🙂

    But Morton, thanks for taking the bait. I actually agree that exploratory search and the ad-supported model aren’t mutually exclusive–in fact, I see the value of the former and the problems with the latter as independent.

    So let’s talk about the ad-supported model. What I don’t understand is how it will persist if it depends on users seeing ads that they don’t want to see. Of course, targeted ads are supposed to be ads that users *do* want to see, but most advocates of the ad-supported model don’t seem to envision a world where it’s easy for users to ignore / skip / block ads. In fact, it seems that there’s a muted arms race between advertisers and ad blockers that only flares up when the ad blockers achieve sufficient popularity. But ads inevitably get annoying enough–in order to make users pay attention to them–that users do block them.

    Basically, if ad targeting were perfect, wouldn’t it be search / information access? Which is to say, wouldn’t the ads actually address user’s information needs so well that a search engine or information provider should have shown the users that information even if the advertisers weren’t paying for the users’ attention? And, if that’s the case, why are the advertisers paying?

  • 8 Morton Swimmer // Mar 26, 2009 at 1:46 am

    Let’s roll up the last point first. There is a difference between finding something and knowing how to buy it. (This is assuming I’ve searched for something that can be bought, but let’s work with this for a while.) I’ve searched and found the camera of my dreams, the Pentacks BC-1. But that search landed me on the Pentacks website, which I can confirm this is what I really really wanted, but didn’t help be actually buy it. However, the search also pointed out a vendor that sells it that matches my preferences (ie, I can pay using PayBuddy in NY-Dollars). So, I buy it, the vendor gladly give the search company a cut, all are happy and flowers turn a lovely shade of yellow just in time to take a picture of them.
    In fact, the manufacturer may also want to pitch in as well, but that would require some backchannel communications between the vendor, the manufacturer and the search engine, which sounds icky.
    So, the vendor is paying the search engine for doing a good job. Why would they pay if the SE is always doing a perfect job? Because there isn’t perfection and there will be very close matches. A vendor can influence how they get positioned amongst the close results and will be willing to pay for that.
    What ads currently try to do often is to get us to purchase something we don’t really want. That wont work any more, but I don’t see that as a reason why ad-supported capitalization will die.
    To address the former point, it is not true that I want to block all advertisement. I just don’t want to see the stuff that is not important to me. If I get zero to, lets say, four ads that are relevant to me and focus my thinking, that is a win-win situation. I just don’t want to see ads for Golf clubs when I was searching for European small cars.
    Cheers, Morton

  • 9 Morton Swimmer // Mar 26, 2009 at 1:52 am

    Hi Jeremy,
    What I am saying is that at the end of the search, there will be still a selection of closely related items or options to purchase and advertisement is one legitimate way of attempting to prioritize the user’s decision. Advertisement is not just pushing produce but also building brands and thereby confidence in a product or vendor.
    Cheers, Morton

  • 10 Daniel Tunkelang // Mar 26, 2009 at 2:16 am

    So, let’s say I’ve decided that I want to buy a specific product. Let’s make it a real camera, e.g., the Sony Black Cyber-shot T500. I suppose I can search for it on Google. There are some ads at the right, which I might turn to because the search results are a jumble. Or I might look at Google Products–which still shows a jumble of natural results and ads at the right.

    But isn’t this the page I want? Note that it doesn’t have ads, and it’s surely the most useful page on Google for someone who has decided to buy a Sony Black Cyber-shot DSC-T500. The only reason I’d click on an ad is if I gave up prematurely en route to such a page.

    I have no problem with a search engine exacting a commission for facilitating sales in which it plays a neutral role. Both the user and the vendor benefit from a more efficient information market. But the ad-supported model is actually at odds with optimizing the user experience. As a user, why would I want the advertisers’ willingness to pay determine what information is presented to me?

    To frame the question differently: ad blocking violates Google’s terms of service, though I’ve never known Google to enforce those terms. If Google thinks the ads are a service to users, then why doesn’t Google–or any other search engine–allow users to legitimately opt out of them? Or, even bolder, make the ads opt-in?

    I think the search engines and advertisers realize that there’s a social contract involved: users are forced to see ads they wouldn’t volunteer for, in exchange for free services. And I think that model isn’t sustainable.

  • 11 jeremy // Mar 26, 2009 at 11:33 am

    Morton,

    A week or two ago, Daniel issued a challenge to find an old cartoon that he was looking for. Read through the comments section here to see how it was found:

    http://thenoisychannel.com/2009/03/14/challenge-blog-twitter-vs-aardvark/#comments

    Basically, nobody found the page, directly. They only found a page that linked to the correct page.

    Similarly, with your Pentacks example, I did a normal, simple web search for “pentax” and, in the second result, found this:

    http://www.pentaximaging.com/

    This is the manufacturer’s website.

    So you say it’s still valuable to have advertising, because I need to find different places to buy it. Well again, right there on the pentaximaging.com web site, in the upper right hand corner (where ads normally are in the Google results, BTW, which means users are already conditioned to looking there 🙂 ) is a link to at least 48 different local and national retailers, a nice handy map interface for me to find local outlets, etc.

    So why do I need an ad to tell me where to buy, especially when Google only offers me 8-11 ads per page. It will take at least 4-5 pages worth of clicking before I see 48 different retailers, compared to the one page, one click that it took, just looking at the manufacturer’s website.

    Google likes to rave about how everything they do is to get the user off the Google site, as quickly as possible, with the fewest number of clicks. Given that most manufacturer’s websites do contain a “Where to Buy” page, I guess I see the Google ads as neutrally useless at best, detrimental to the user’s search at worst. If you were to rely on Google ads to find 48 different retailer options, Google would force you to do a heckuva lot of work.

    (And yeah, I also want to echo Daniel’s question: Why would I buy from a retailer that had a larger advertising cost? I’d rather buy from a retailer that didn’t have to pay the advertising overhead, and thus could pass the savings directly along to me.)

    (By the way, thanks for being willing to engage 🙂 )

  • 12 Carl Grimm // Mar 26, 2009 at 3:13 pm

    Time to ring in Gentlemen; Daniel I always love playing the Devil’s Advocate, perhaps a side effect from years on the collegiate debate team. Count me in and here we go wtih this two part post.

    Let us be pragmatic with respect to the state of Internet wide search. From a business standpoint there is a demand for the ability to search. This demand constitutes a market and the market yields certain demands. There is a real cost associated with supplying search to this market.

    Any businessperson wishing to make a profit will focus in on the segment of the market that is most profitable and service their needs. The business problem that immediately creeps up is in measuring the value to the consumer of providing the service. Consumers, thanks to television and radio advertising, have been spoiled over the last few decades into thinking they do not have to think about the value of things they receive. It is very easy however for a person selling a product or service to understand advertising ROI and generally know exactly how effective it is and how much they should pay.

    The subtlety that most search geeks fail to see is that Google does not fulfill the consumer’s demand for search; search is a means to an end, a conduit to fulfilling a prime need. Because Google focuses on the mainstream consumer to capture their value, most people want to find the most popular things. Google makes a bet that you are in essence an average consumer.

    Google get’s paid by serving ads that may assist you in fulfilling your prime need. If I am looking for cameras the ads may help me get there whereas if I am looking for a specific camera the ads may offer an opportunity for a specific vendor to advertise that they sell it or for a competitor to sell you a different brand of camera. My prime need it to take a picture. The greater and very well known value of this transaction lies on the side of the vendor. Google therefore evolves a model that supports these kinds of transactions. They are simply serving their market.

    To answer your question, “Why do the advertisers pay?” They pay for a click, which is a shot at capturing the some of the value that is in fulfilling the consumer’s prime need. Products themselves aim at fulfilling these prime needs. My need is to take a picture. I may be under the belief that a Sony Cyber Shot 500 fulfills this need but marketers know there is an opportunity up until the point of purchase that they can offer a different product that fulfills their need.

    Morton this is where I disagree with you on the statement that ads serve to get people to buy things they don’t want. Customers may buy things they don’t need but I maintain they never buy something they don’t want.

    Search engines are not omniscient and users are traditionally very bad at communicating their need in a search. You point to the specific page you think that Google should show you if you want to purchase the Sony Black Cyber-shot T500. It will if you tell it you are shopping for something by using the product search and specifically tell it that is the exact camera you want by selecting it from their “catalogue” of items. However your search in the general Google shopping site also matches the Sony Cyber-shot T500 Canon Digital Camera Black Case. Google only has to work with what you gave it. Interestingly enough if you put in quotes “Sony Black Cyber-shot T500” Google products returns zero results. It is not “smart” enough to figure out what you the ends user wants just from those words. They know that what people type in is more often than not only part of the picture and they refrain from making too large a guess too quickly.

  • 13 Carl Grimm // Mar 26, 2009 at 3:14 pm

    If the user experience revolves around fulfilling the need to take a picture, what I am calling the prime need, then an ad that leads me somewhere is absolutely not “ad odds with optimizing the user experience.” It may very well indeed lead to the consumer purchasing a camera that meets their prime need more so than the original one they had in mind because they saw it in the paper or a friend told them about it. There have been many cases where I am looking for alternatives to an item and very much know that if I just type in what I am looking for the fastest way to find alternatives is to look at the ads.

    You speak of this model acting against an “efficient information market.” In a system where the economic value can only be measured at one end, the end of the vendor of the good or service, the ad supported model is the winner because it is the sole determinant of value. Google cannot ask the user before searching how much it is worth to them to show them other products that may fulfill their prime need or ask for a cut of the savings if they find the camera 15% less at another vendor.

    The ad supported model of Internet search is currently the only sustainable model at this point. Why? Because up until now is the only model of sustainability that the consumer demands. I am a firm believer efficient markets and if there was a foreseeable profit to be made in ad free Internet search there would be a firm on top of it.

    We live in a world where there is only still a handful of advertising free premium content providers like HBO. The interesting thing about that industry is the advertising model has to be reconsidered not because the HBOs of the world posed a threat to the advertising model, the DVR and Internet distribution posed the threat. I believe it will be the same with Internet search. The ad supported model is here to say for a very long time until something else comes along to destroy it. That destroyer will not be paid ad-free search however.

  • 14 jeremy // Mar 26, 2009 at 5:19 pm

    Carl, thanks for the thoughtful, detailed response.

    Let me quickly comment on just one portion of what you said:

    If the user experience revolves around fulfilling the need to take a picture, what I am calling the prime need, then an ad that leads me somewhere is absolutely not “ad odds with optimizing the user experience.”

    I will grant you that the ad might not be at odds with fulfilling the user information need (i.e. the ad is relevant), but what I question is the optimization.

    What if, instead of showing an ad over on the right hand side (the first place the user looks other than the organic results), the search engine were to show a link to Google Product Search Results, pricegrabber.com, shopzilla.com, Yahoo shopping, or any of a number of dozens of price comparison and shopping locations? Might any one of those links not be better, overall, to the user experience than just a couple of advertising links from just a couple of vendors?

    My point is that the more information provided by a shopping comparison site or a shopping vertical is of much more use to the user than a few ads. So ads, while not completely irrelevant, are not optimal user experience, and Google does its users a disservice by showing the ad, rather than showing the shopping/comparison information.

    There have been many cases where I am looking for alternatives to an item and very much know that if I just type in what I am looking for the fastest way to find alternatives is to look at the ads.

    Oh, this is a great example, Carl! What you are describing here is a perfect exploratory search scenario. Go to Kosmix.com (a site that I am, like Daniel, also not affiliated with) and type the query [canon g9]. The G9 is a camera that I would like to someday own.

    http://www.kosmix.com/topic/canon_g9

    Look over on the right hand side. In addition to finding specific, related digital camera suggestions/recommendations, it also provides information on different camera category types (DSLR vs. point and shoot), different features offered by digital cameras (image stabilization, shutter lag, lens hoods) and various cameras that have the live preview feature. Is the set of information perfect? No. Is it much more information than I could have gotten from a few ads? 250% yes.

    Sponsored results (ads) still exist on Kosmix. But you have to scroll down to get to them, and they’re buried off to the far lower right hand side of the page.

    In short, the ads are “optimized” w/r/t the amount of incremental information that they offer to the user, which is to say that they offer far less information than an exploratory set of facets, categories, and other related options. All this exploratory information takes its rightful place on the top right hand side of the page, where it belongs.

    I have not addressed the rest of your comments, which are about the feasibility of ad-supported vs. non-ad supported models. It may very well be that no one will pay. But what I find absolutely disingenuous is when people claim that the advertisement is the optimal piece of information to fill that right hand side portion of real estate on the results page. Kosmix clearly demonstrates that there is a lot more valuable information that you can offer the user, than an ad, when it comes to what you call the ‘prime user need of taking a picture’.

  • 15 jeremy // Mar 26, 2009 at 5:34 pm

    Sorry, I mistyped something above in comment #14. What I meant to say was:

    In short, the ads are NOT “optimized” w/r/t the amount of incremental information that they offer to the user, which is to say that they offer far less information than an exploratory set of facets, categories, and other related options. All this exploratory information takes its rightful place on the top right hand side of the page, where it belongs.

    Go to Google, and type the “canon g9” query. You’ll get shopping ads for the G9 and the G10. Go to Kaltix and type that same query, and you’ll get not only more information about the G9, but you’ll get suggestions which include the Powershot A650, Powershot S5, Powershot SD870, the Powershot S3, the Canon 40D, the DMC-FZ18, the Nikon Coolpix, and the Nikon D3.

    So if your primary information need when you type [canon g9] is to find out more information about the G9, Kaltix does a much better job that Google. You can see everything from the home page of the camera itself, to pictures that were taken with the G9 by average users, to comments and blogs that talk about the G9.

    However, if your primary information need when you typed [canon g9] was to find any instrument for taking pictures, and not specifically the G9, then Kaltix’s exploratory search also does much better than Google’s ads, because you get much more information on a larger variety of other cameras from a number of different manufacturers. All without relying on advertising to do so!

    (BTW, if your primary information need really is to find any instrument for taking pictures, why you typed [canon g9] as your query instead of [digital camera] is kind of a mystery to me. But so be it. Kosmix still does a better job at that primary need, with the query [canon g9], than does Google.)

    So the way it appears to me is that exploratory search is ultimately more relevant to the user’s information need, than an advertisement.

    For the other reasons you list, Google might still promote the ads above all else. But it is absurd for them to continue to say with a straight face that doing so is the optimal or best use of all available information. Ads provide an overall worse quality experience for the user, because they substitute or replace information that could have been better.

  • 16 Carl Grimm // Mar 26, 2009 at 6:23 pm

    Jeremy,
    I am assuming that Google places ads in a way that maximizes the revenue from their advertisers while balancing user expectations and usability; a result that we would expect from any rational for-profit corporation. I am saying that the ad placement is optimal from Google’s standpoint, which means optimal to ensure they can pay the bills for the server farm. It is the tradeoff we accept for free search; there is a definite measurable trade of value going on here. Google after all is an advertising company who owns a search engine. We expect that the ad on the billboard is from whoever paid, not the best restaurant off the exit.
    I trust the ads on Google for exploratory means in certain cases because I know a human being put up some cash and actually thought about being able to fulfill my prime need rather than some unsupervised clustering or related topic extraction engine taking a guess and failing miserably.

    This “helpfulness” gets them in trouble all the time. They tend to gush information that Google keeps tamely a click away. For queries like “Hilton Hotel Paris” Kosmix brings fills up your screen real estate with videos and blogs tweets related to Paris Hilton. Ironically they also pull an Answerbag question, “Why is it I can’t search for information on Hilton Hotels without digging thought pages of crap about Paris Hilton?”
    Kosmix utterly blows up with useless information for this query. The exploratory information in the upper right has 1 link out of 17 that actually have to do with the hotel. I simply cannot see how this overall yields a more optimal model for information discovery. Google nails this query in my mind and I would rather deal with an extra click or two or to refine my query than to be knee deep in irrelevant information

  • 17 Digvijay Lamba // Mar 26, 2009 at 7:20 pm

    I am an engineer at Kosmix and have been following your discussion with a lot of interest.

    I want to specifically address the “Hilton Hotel Paris” and “Paris Hilton” case.

    Please note that the fact that Kosmix does not do well for this query is a strike against us at Kosmix and not against exploratory search in general. I believe the model is fundamentally sound, even though we need to do a better job at it at Kosmix.

    Distinguishing between Paris Hilton and Hilton Hotel in Paris should be something we do, (Like we disambiguate several other queries), and we are fixing these problems at a rapid pace.

    However, I believe, as we correctly interpret more and more tricky queries we will start enabling users to explore those tricky topics in completely different ways than traditional search allows today.

    For example look at the page for http://www.kosmix.com/topic/Paris_Hotels. It still allows a lot of rich connections for someone planning to travel to Paris.

  • 18 jeremy // Mar 26, 2009 at 7:55 pm

    I am assuming that Google places ads in a way that maximizes the revenue from their advertisers while balancing user expectations and usability; a result that we would expect from any rational for-profit corporation…It is the tradeoff we accept for free search; there is a definite measurable trade of value going on here. Google after all is an advertising company who owns a search engine. We expect that the ad on the billboard is from whoever paid, not the best restaurant off the exit.

    So let me clearly and succinctly restate what you are saying: Google provides somewhat of a less than optimal experience for its users, in order to make money from its advertisers. Is this a fair and truthful statement? Please, correct me if I am wrong.

    Because if it is correct, let me juxtapose what Google actually says in their “10 Things” corporate philosophy with what we here have collectively determined is the honest truth about their corporate philosophy. This should be fun.

    http://www.google.com/corporate/tenthings.html

    Google: From its inception, Google has focused on providing the best user experience possible.

    Our Conversation: From the inception of the first advertisement that Google showed on its site, Google has focused not on providing the best user experience possible, but the best user experience possible within the limitations of making sure that we make a profit

    Google: While many companies claim to put their customers first, few are able to resist the temptation to make small sacrifices to increase shareholder value.

    Our Conversation: Actually, we at Google have made small sacrifices. There is better, more useful, more relevant information that we could have shown you, the user. But we don’t, because we need to get paid. Now, please trust and believe us that the compromises and sacrifices that we have made are not malicious. We would still like to see you, the user, succeed. But we’re not going to help you as much as we otherwise could have, because by so doing, you would click fewer ads, and we wouldn’t get paid. But we’re not evil. Trust us!

    Google: Google has steadfastly refused to make any change that does not offer a benefit to the users who come to the site:

    Our Conversation: Google has steadfastly refused to make any change that does not offer a benefit to the users who come to the site. But conversely, we also steadfastly refuse to make any changes that offer too much benefit to the users who come to the site. That is, we will make a change to the site if and only if it simultaneously does not harm our users, but also does not harm our advertisers. Unfortunately, this means that there are some positive changes that we could have made that would benefit the users even more, but we won’t make them because they harm the advertisers

    Right? This is an honest statement of Google’s official position, in as far as I understand it from what you’ve told me. Correct?

  • 19 Carl Grimm // Mar 26, 2009 at 8:11 pm

    Digvigay, Thanks for chiming in (or stepping into the ring depending on how things go.)

    I am going to have to ding you on this page as well. Assume I am actually looking for a hotel in Paris. I have to literally go an entire page down on my monitor (set at 1600 x 1200) to get the name of a single hotel in Paris. The first two bits in “Travel Articles” are for the Paris Las Vegas. If I actually wanted to book at hotel you have only pointed me to a bunch of other sites with the information I need and wasted time.

    Google would have given me 10 hotels actually in Paris France already and shown me them on a map rather than showing me little pictures of hotel lobbies of which I have no clue where they are. I now know about a few places I can shop in Paris and a few of the districts thanks to the related section yet I no absolutely nothing about Paris Hotels without wading way below the fold.

    The only result that may be what I am looking for is the George V Hotel and it comes up beneath a bit from Historic Hotels in New Orleans. Perhaps I’m taking my Devil’s Advocacy too far but I literally think the results are rubbish.

    In fairness your topic page on just Paris is much better, offering general information, tourist attractions, landmarks, etc. If I wanted general information on Paris it would be a nice page to visit to start my journey.

  • 20 jeremy // Mar 26, 2009 at 9:02 pm

    For what it’s worth, gang, I just tried the query [paris hilton] on Google, and got a results page with 15 results filled with nothing but the socialite. No mention of the city, or any hotels, anywhere.

    The Google advertisements that I am seeing as a result of this query are for perfumes, shoes, and gossip rags. Not a single hotel advertisement. For any hotel chain in any city.

    I thought to myself: Uh oh.. maybe that’s because I’m logged in, and doing a personalized search. So I logged out of Google, and I’m getting the exact same results.

    So what are you doing differently, Carl, that your Google search using the same query [paris hilton] is getting you the map and the hotels? Please be specific.

    I just tried another query [hilton paris]. That gave me the map and the 10 hotel links.

    So Google appears to be query term-order sensitive, and Kosmix does not. To me, that is not an indictment on Exploratory Search, as Digvijay points out. It is an imminently and trivially fixable shortcoming of Kosmix.

    Remember, Google suffered from the same socialite error, when I entered the query terms in that [paris hilton] order.

    If you really want to evaluate Kosmix’s exploratory search, try [london hilton] instead:

    http://www.kosmix.com/topic/london_hilton

    When I do that query, as an honest attempt to evaluate the quality of Kosmix’s exploratory search, rather than the quality of their query disambiguation, a different picture emerges. Look over on the right, and I see (1) a list of 3 hilton hotels in London and (2) a list of 6 other non-Hilton hotels in London.

    The Hilton London hotels are clearly separated from the Non-Hilton London hotels. After all I did specifically as for Hilton, and so Kosmix respected that query request. But it simultaneously gave me exploratory options, and let me see other hotels as well — but clearly separated from the Hilton ones.

    Google on the other hand gives me 10 Hilton hotels, only. I admit, the map interface is nicer than anything that Kosmix has. Score one point for Google on that one.

    But Carl, you said earlier that your primary information need when searching for [pentax camera] was on finding any device to take pictures.. and you liked the fact that advertisers provided links to non-pentax information. So I assume that you would be much more pleased with Kosmix’s listing of non-Hilton London hotels. Score one point for Kosmix.

    On the query [London hotel], I would say that Google and Kosmix appear to be equally scored.. 1 to 1.

    We would have to run a lot more queries, though, to do a more robust comparison.

  • 21 Carl Grimm // Mar 26, 2009 at 9:11 pm

    Jeremy,
    Now we are wading into the land of corporate philosophies. Who knows if Google is a bunch of realists or existentialists? I seek not to criticize their actions based on how I believe they act according to my interpretation of their philosophies. They may very well believe that their search results are superior and get the user to the most relevant information the fastest and with the best user experience.

    I take little stock in phrases like “the best X possible” as it is extremely relative and without a discussion that wanders into epistemology it really is meaningless.

    If we personally believe that Google is not serving our search needs we have the ability to “take our business” elsewhere. I do not fault any corporation for providing the best possible user experience possible within the limitations of making a profit. A corporation is made to make a return on investment creating value. Google is beholden to its users only because they must please them to make a return.

    Their goal is to maximize profits while maximizing user experience. If they drop user experience while chasing profits, they will only do so based on the elasticity of their user base. If a corresponding drop in “user experience” based on how you see it causes an increase in profits Google will presumably do it until they chase enough people away to equal out the gain. Likewise if it costs them $1 to make an improvement to “user experience” but it only yields them 90 cents in return why would you expect them to waste shareholder value to create value for the users if the users don’t pay?
    We as rational consumers must be expected to maximize our value and Google as a rational for-profit corporation must be expected to maximize its value. Because we do not directly cut a check to Google when utilizing its services the only way we can factor into the equation is to walk away.

    The beauty of the market is there are many providers. The market allocates resources based on the maximization of value to all stakeholders. Because the definition of optimum user experience is subjective many providers will always exists to cover the variances.

    You clearly believe that Google delivers a “less than optimal” experience. If it is generating a profit then you have to assume it is fulfilling the needs of its target market. My questions to you Jeremy is are you upset that Google is leaving money on the table because it would gain more profit by maximizing user experience or are you upset that is makes so much money and it is sub-optimal based on your standards?

  • 22 Carl Grimm // Mar 26, 2009 at 9:19 pm

    But I wanted the Hilton hotels in London, that’s what I actually typed.

    The results set does not even include a link or anything to the London Hilton!! Google gives me 7 Hiltons in the London area with links and phone numbers.

    I don’t mind being told about other hotels but if you fail to find what I am looking for it is an utter failure. Google pulls all Hiltons in London while Kosmix pulls me a snippet of a Wikipedia article, some pictures of a building and some videos that are barely related. Fashion week in LA anyone?

    Do I have to beg for a website link to Hilton and some phone numbers? Kosmix is clearly more experimental than it is search because it does not find anything I am looking for.

  • 23 Carl Grimm // Mar 26, 2009 at 9:53 pm

    Let us assume my prime need is indeed to stay at a hotel in London. I do desire that you actually help me find one.

    So while Kosmix pulls up the Savoy Hotel up in the right and it catches my attention because I’ve had a Savoy cocktail before. Fulfilling my need of actually contacting a hotel in order to fulfill my primary need is a very important part of the process.

    So Kosmix, you got me, your exploratory search function has convinced me the Savoy is the place for me. Where is the website or phone number? I actually have to click on the Wikipedia page and hope it has the website to get to the phone number! Kosmix to Wikipedia to the Savoy’s website. Yikes. Google had a phone number right there.

    If I just wanted to learn about hotels around London and their histories by having it point me to Wikipedia pages and related hotels Kosmix would be great.

    Problem is most people are trying to book a hotel and not write a research paper on them.

  • 24 Daniel Tunkelang // Mar 26, 2009 at 10:11 pm

    Carl, is this a reasonable synopsis of your argument in favor of the ad-supported search model?

    1) Consumers expect search to be free.

    2) Sellers are willing to pay for advertising because they see the ROI.

    3) Google is win-win for consumers and sellers because it provides ROI for sells and leads consumers to the popular products they want.

    4) Advertisers are competing within the fungibility of options consumers don’t care about–that’s why Google can make them bid for the privilege without detriment to users.

    5) Google is thus giving users a satisificing experience–good enough fulfillment of their information needs, at a great price.

  • 25 Daniel Tunkelang // Mar 26, 2009 at 10:30 pm

    Digvijay, thank you for engaging the conversation here, and props for having the courage to admit where Kosmix falls short.

    I think it’s safe to say that Google is still the winner when the query so clearly expresses a specific information need satisfied by a single document. I don’t think Google’s rivals are far behind, but I’m ready to crown Google the winner for known-item search.

    On the other hand, I think that Google is neglecting the basic support for exploratory search, and that Kosmix is probably the best game in town today, at least for the open web.

    But here’s what’s interesting. When I have a known-item search, I actually am happy to bypass the Google results page, using the I Feel Lucky interface by entering my query into the Firefox address bar. When I don’t, I’m more likely to turn to Kosmix, even if it is slower and quirkier than Google.

    So, in effect, I use Kosmix more than I use Google. It’s worth noting that Kosmix usually include top results from Google. But I believe those are pretty interchangeable with top results from Yahoo or Live.

  • 26 jeremy // Mar 26, 2009 at 11:36 pm

    Now we are wading into the land of corporate philosophies.

    Only because Google makes it such a point to publicly and constantly proclaim how different and wonderful they are. If they never made such a big deal about it, themselves, neither would I.

    Who knows if Google is a bunch of realists or existentialists?

    I have no idea either, and it’s not relevant to this discussion.

    I seek not to criticize their actions based on how I believe they act according to my interpretation of their philosophies.

    Nor do I. Rather, I seek to criticize their actions based on their own interpretation of their own philosophies. Let me point you back to the Google paper that started it all:

    http://infolab.stanford.edu/pub/papers/google.pdf

    Quoting from that paper:

    Currently, the predominant business model for commercial search engines is advertising. The goals of the advertising business model do not always correspond to providing quality search to users.

    and

    In general, it could be argued from the consumer point of view that the better the search engine is, the fewer advertisements will be needed for the consumer to find what they want. This of course erodes the advertising supported business model of the existing search engines. However, there will always be money from advertisers who want a customer to switch products, or have something that is genuinely new. But we believe the issue of advertising causes enough mixed incentives that it is crucial to have a competitive search engine that is transparent and in the academic realm.

    Does Google support itself via ads? Yes. Is Google transparent about how it ranks those ads, when and why it places ads above the organic search results, in addition to alongside them? Is Google, overall, transparent and in the academic realm? No. Therefore, they do not live up to their own standards. QED.

    My questions to you Jeremy is are you upset that Google is leaving money on the table because it would gain more profit by maximizing user experience or are you upset that is makes so much money and it is sub-optimal based on your standards?

    I’m upset with neither of those things, because those are both questions about money. I am upset because they do not provide the best user experience possible… despite claiming to. That’s it.

    I absolutely agree with you that it is any companies right to do whatever it wants, within the law. I have no problem with them doing so, either. What I have a problem with is them providing search experiences that are non-optimal, but saying that they do otherwise.

    Consistency. That’s all this is about. No more, no less.

    Change the policy/mission statement, or change the implementation. Either way in totally fine with me.

  • 27 jeremy // Mar 26, 2009 at 11:45 pm

    Carl, in response to the [London Hotel] query, let me just point out that Kosmix did not fail. Kosmix lists 3 Hilton hotels at the very top of the search results page:

    Hilton Garden Inn
    Hilton London Paddington
    Waldorf Hilton

    You’re not used to looking at the right hand side of the results first, rather than second, but there is your answer, right at the top, with no ads above it.

    So before you say that a query fails, look a little closer.

    That Google gives you phone numbers on the results page is already something I gave Google a point for. I agree, that’s more useful.

    But I have a serious problem with your response in comment #22. More on that in a second:

  • 28 jeremy // Mar 27, 2009 at 12:00 am

    Ok, Carl, explain this one to me.

    In comment #22, when you said you asked for Hiltons in London, you got upset at Kosmix, because it returned other hotels in London that were not Hiltons. Right? Here are your words:

    But I wanted the Hilton hotels in London, that’s what I actually typed.

    However, when we go back up to comment #12, you write:

    “If I am looking for cameras the ads may help me get there whereas if I am looking for a specific camera the ads may offer an opportunity for a specific vendor to advertise that they sell it or for a competitor to sell you a different brand of camera. My prime need it to take a picture……Products themselves aim at fulfilling these prime needs. My need is to take a picture. I may be under the belief that a Sony Cyber Shot 500 fulfills this need but marketers know there is an opportunity up until the point of purchase that they can offer a different product that fulfills their need.”

    So here you seem to be saying that the advertising on a Google results page is still relevant, because your primary/fundamental need is to take a picture. You say that you might be laboring under the false impression that the Sony Cybershot is the tool you want to take that picture, but that Google still provides a fundamentally relevant experience because it’s showing you cameras from other vendors, which cameras are also fundamentally capable of taking pictures.

    So why did you not instead complain that [Sony Cybershot] is what you typed, and fault Google’s ads for not being exactly relevant to what you wanted?

    Why do you instead fault Kosmix when it shows you hotels in London other than 3 Hiltons? Can’t I make the exact same argument? Can’t I say that your primary need is to find a place to lodge in London during your stay, and that Kaltix, by showing you other hotels in London, is doing just as good of a job as those Google ads that show cameras other than the Cybershot?

    This feels like a big, big inconsistency to me. It feels like you’re defending Google just because it’s Google.

    So which way is it? If Kosmix is bad in the hotel case, then Google’s ads are bad in the camera case (because they’re non-relevant to “what I actually typed”), and therefore Google is violating its self-stated principle to only show ads if they’re relevant.

    On the other hand, if Google’s ads in the camera case are relevant, then you have to admit that Kosmix’s other London hotel exploratory suggestions are also relevant. And in that case, Kosmix manages to give you those suggestions, without advertising — something the Google founders themselves argued stays truer to the needs of the user, rather than the needs of the advertiser.

    So, what’s the story?

    I’m really not trying to turn this into a flame war, by the way. If it starts to feel that way to you, we should stop.

    I just see Google failing its own standards. Not my standards, but its own. And that’s why I agree with Daniel about the value (rather, the lack thereof) of the advertising-supported model.

  • 29 jeremy // Mar 27, 2009 at 12:05 am

    Can’t I say that your primary need is to find a place to lodge in London during your stay, and that Kaltix, by showing you other hotels in London, is doing just as good of a job as those Google ads that show cameras other than the Cybershot?

    Sorry, I mean Kosmix, not Kaltix 😉 Sorry, Kosmix.

    But again, Carl, why do you say that Kosmix is bad when it shows you other London hotels, even though your primary need is finding a place in London to stay, but praise Google when it shows Nikon ads in response to your [sony cybershot] “what you typed” query?

    It doesn’t make sense.

  • 30 jeremy // Mar 27, 2009 at 12:11 am

    And you still haven’t answered me, from your Comment #19, about how you managed to get Google to show you a list of hotels for [paris hilton]. When I tried the exact same query on Google, I got the socialite, and ads about perfume and shoes. Not a single hotel number.

    What did you do different?

  • 31 Ross Bates // Mar 27, 2009 at 12:59 am

    Daniel – you question the value and longevity of the ad-supported search model, but I’m wondering what you propose as an alternative. Let’s suppose that google.com redirected to endeca.com for the next 4 weeks. You have hundreds of millions of visitors coming to your site looking for information on the public internet – who pays for your costs?

  • 32 Carl Grimm // Mar 27, 2009 at 1:14 am

    In response to #30 Google returns very different results for the query “Hilton Paris” than for “Paris Hilton.” The former will trigger a results list that contains the map and business results and the the latter assumes you are looking for the person.

    Kosmix per comment 17 by does not treat the two as different.

  • 33 Carl Grimm // Mar 27, 2009 at 2:06 am

    In response to 27 and 28: With Kosmix I do look at the upper right hand “results.”

    Why I consider Kosmix a failure is that is returns rather horrid actual results for Hilton London. This has nothing to do with the exploratory function.

    Kosmix returns as what I consider the prime result for that query a single Hilton in London, the Wikipedia entry for the Hilton hotel on Park Lane. No phone number and no web site. What is it assuming I was looking up the hotel in the first place for? Hopefully not to actually do anything other than read about them.

    Kosmix also returns results in the “Related in the Kosmos” results to 4 other Hiltons in London.

    If I am looking for all the Hilton’s in London, why do I have to further tell the engine to enumerate on what I asked for in the first place?

    I am by no means saying the Kosmix returning a list of other hotels in London is bad or worse than Google Ads. As far as the exploratory function that that specific piece serves I will concede in this case is superior to Google, who will only show me ads that humans have deemed related to my query per their hope to make money from me.

    However, I do not let Kosmix off of the hook for returning a Wikipedia article related to a single hotel along with irrelevant pictures and videos. The fact that it finds other Hilton in London as “related” is very poor. Other Hilton hotels in london are not related, they are what I asked for in my query.

    I suppose what I am saying is that I will not tolerate what I consider a sub standard primary search function for a a better exploratory search.

    Why I do not specifically call out Google’s ads as being “bad” in the camera case is that I know the ads for what they are, people who are bidding on my keyword. They are not exploratory search from an intended standpoint. I may however leverage them as such.

  • 34 Carl Grimm // Mar 27, 2009 at 2:07 am

    In order for exploratory search to take off it must go above and beyond expectations. When I type in “Hilton London” I expect that all Hilton hotels in London will appear right at the top. Not one here and four over to the right as related.

    I agree that Google is not living up to its philosophy per my view either. However, I make allowance for the fact that they may actually believe they adhere to it and as thus it does not bother me. “Actions speak louder than words,” as my mother often told me. I take what they actually do as the execution of their true philosophy. Perhaps I am too pragmatic.

    If you think I am defending Google for Google sake pop over to enterprisesearchblog.com and read my entry “Why Wikia Search must Prevail.”

    Just because Google does not live up to their self proclaimed standards does not make the advertising supported model not viable. General Internet engines will never in my opinion be able to charge the user for search and thus will always have to rely on a sponsored model.

    By delivering a user experience to access public information, unlike LexisNexis which can charge for content unavailable anywhere else, the Internet engines are doomed by the fact that users do not expect to pay as they cannot effectively monetize their own search use cases due to the general media paradigm that is pervasive.

    We live in an era were we have newspapers shutting down because people will not pay for content. What makes us think they will pay to find the content?

    Winston Churchill said, ” Democracy is the worst form of government except for all those others that have been tried.” I think most of us may very well agree that Google is the worst Internet search except for all those others that have been tried.

  • 35 Digvijay Lamba // Mar 27, 2009 at 4:03 am

    First I want to thank everyone for the feedback and for pointing out what are, obviously, some issues at our end.

    At Kosmix, we launched our product a few months ago on the “Browse” or “Explore” model. If you were looking for a lot of information on a topic, we aim to present it in a well organized fashion on the page. We have not really focussed enough on queries that look for the needle in the haystack and have one right answer. You would probably just Google those.

    But if you were not sure which hotel to stay at or what attractions to visit on your trip, Kosmix aims to provide you with a comprehensive guide of what’s available on the web.

    However, I think we have all realized that our initial focus, that of exploration, is only one element of exploratory search. We at Kosmix need to do much better at answering the query itself when the user’s intent is clear, and then providing richer opportunities to explore. We do include Google Search Results but, clearly, we need to do a better job.

    Fundamentally, there is no reason why “exploratory” or “semantic” search cannot both answer your question and connect you to other related rich information that you would be interested in. It is a hard problem, but I believe we have taken a good first step.

  • 36 Does Metadata Matter? | The Noisy Channel // Mar 27, 2009 at 9:54 am

    […] trilled at the discussion that my call for devil’s advocacy has incited. Keep bringing it–and let me know if you’d like to contribute a guest […]

  • 37 jeremy // Mar 27, 2009 at 11:28 am

    In response to #30 Google returns very different results for the query “Hilton Paris” than for “Paris Hilton.”

    Ok, I just wanted to make sure. That’s what I tried as well, and got those different results.

    Just note, however, that for the [paris hilton], your original query order, before you had to retype it a second time, Google was absolutely no better, than Kosmix — and had even further non-relevant ads for perfume and shoes.

    Why I consider Kosmix a failure is that is returns rather horrid actual results for Hilton London. This has nothing to do with the exploratory function.

    Actually, Carl, let me submit to you the notion that is does have everything to do with the exploratory function. You’re familiar with Andrei Broder’s 2002 paper, aren’t you? In that, he lays out a taxonomy of web searches. He says that there are 3 types: navigational (home page / known item finding), transactional, and informational.

    Google is optimized toward navigational and transactional searches. And you as a searcher are conditioned toward expecting those types of results from Google. That’s why you expected to see phone numbers in the results list, for your [london hotel] query.

    So Google does better on transactional queries. But Kosmix is optimized for, and does better on, informational queries. Google is terrible on informational queries.

    For example, suppose you had just moved to Canada, and discovered the sport of curling. You didn’t know much about it, and wanted to learn what it was all about. You weren’t looking for a phone number to book curling lessons, you were looking for information on the topic. Compare Google:

    tp://www.google.com/search?hl=en&q=curling&btnG=Google+Search&aq=f&oq=

    ..with Kosmix..

    http://www.kosmix.com/topic/curling?

    Or suppose you were a software developer, and interested in learning more about probabilistic graphical models, to use them to help solve some problem or product that you were working on. Compare the query [Markov Random Fields] on Google and Kosmix:

    http://www.google.com/search?hl=en&q=markov+random+fields&btnG=Search

    http://www.kosmix.com/topic/markov_random_fields?

    Kosmix wins hands down, any day. The “at a glance” is much more informative / useful. The “hot in the kosmos” area gives me loads of related concepts, other graphical models, etc. If I had an “informational” information need, I would go to Kosmix over Google, any day.

    So when you say: “I suppose what I am saying is that I will not tolerate what I consider a sub standard primary search function for a a better exploratory search.“, I think what you mean is that you don’t want to use exploratory search for all your navigational and transactional queries. I completely agree with you. Exploratory search has its rightful time and place. By the same token, however, I would also not want to only have the ability to use a navigational/transactional engine (Google) when my information need is “informational”.

    So my ongoing, primary point is that part of what Google really needs to do is develop and give users the ability to switch to exploratory mode, when they have that type of query. You can still make the default navigational/transactional. But my ongoing contention is that by having an ad-supported business model, which always requires ads to appear to the right of the results, Google is fundamentally incapable of ever delivering any results as information-rich and exploratory as Kosmix. And by being so limited, they are doing their users a true disservice, and making pretty large sacrifices to place the needs of the advertisers above the needs of the users. I know you disagree with me, but I just wanted to bring this thread back to that original starting point, to tie it all together.

    Good chattin’ with ya! 🙂

  • 38 jeremy // Mar 27, 2009 at 11:37 am

    One more comment:

    What I dislike about Google is that it encourages and promotes laziness. A good information professional, a good citizen of any democracy, a good consumer, should not rely wholely on any one tool for their answers, prices, information.

    By Google, users have been conditioned to enter all their queries into Google an expect “the right answer”. What they fail to understand is that, depending on the type of question they are asking, some questions might be better handled by Google, while other questions are better handled by Kosmix. To the extent that Google doesn’t help users find Kosmix, I also consider that a failure.

    What I wonder is if you were really interested in [london hilton] (or [hilton paris] or whatever), why you just didn’t go to hotels.com. At that site, you’re not dependent only on the top 10 results that Google gives you. You instead have a whole host of specific options for refactoring your search by hotel name (e.g. hilton), guest rating, star rating, price range, available amenities, property type (hotel vs. suite vs. b&b, etc.)

    This is all about knowing what the right tool is for the job. If you’re just looking for someone’s homepage, or a quick phone number lookup, Google is that tool. If you’re looking for information on a new (to you) topic, Kosmix is that tool. If you’re looking to book a hotel in London or Paris or Boise, hotels.com is that tool.

    You wouldn’t use a screwdriver to pound in a nail. You would use a hammer.

  • 39 jeremy // Mar 27, 2009 at 11:46 am

    Just because Google does not live up to their self proclaimed standards does not make the advertising supported model not viable. General Internet engines will never in my opinion be able to charge the user for search and thus will always have to rely on a sponsored model.

    Two questions:

    (1) What if Google started charging tomorrow? Would you be able to live without it? I would. To me, that says a lot.

    (2) Conversely, would Google ever go for the freemium model? Give it away for free, with ads, for those who don’t care about mediocre information. But then open up and allow (not force, but allow!) other users to pay you for extra, exploratory, informational, useful service. I consider it a real indictment against Google that they don’t allow me to pay, even if I wanted to. And that’s part of the reason why I believe that the ad-supported model is not good, untenable, in the long run. Google is completely unwilling to entertain the idea of providing a service to me that is free of the biases and limitations imposed upon it by its advertisers. Google therefore chooses to stay in the lowest common denominator category, giving trivia answers to trivia questions, and not encouraging or enabling any deeper, exploratory, informational thought.

    In my mind, Google is not that different from Fox News. Keep it simple.. and dumb.

    If that’s the world that you want to live in, go right ahead. But I will cheer anyone on who strives for more.

  • 40 Carl Grimm // Mar 27, 2009 at 2:36 pm

    Jeremey,
    I very much so agree that users are conditioned to only look in one place, for most people that is Google, and come to the conclusion that if they cannot find it on Google that is not there. Now to be fair I am not so entirely sure that Google actually encourages or promotes this laziness. To me this is a research behavior problem compounded with general human behavior. I am lucky that my high school actually had a research skills class as a requirement my freshman year.

    I am familiar with the different kinds of searches per Broder’s definition. I do wholeheartedly agree that you need to use different engines for different kinds of searches.

    You are correct that if I were looking to find hotels in London that I would skip Google completely, I use travel specific websites like Travelocity and Priceline when I have transactional goals to achieve.

    Personally it would be a hard adjustment without Google since they power even the actual part of Kosmix that gets you to the content in the end.

    Here is my final take on it. I concede that Kosmix beats out Google for purely exploratory search.

    I suppose the final question is why Google has not tackled this space? Google in my mind has beat out other engines by providing what most people want. As I said very early on, Google assumes you are an average person looking for average things. I am also unsettled by this and absolutely side with you in that good information professionals and good citizens of any democracy need quality information discovery tools and habits.

    Google is a business like any other and so I really can only fault them for leaving value on the table. I do hold them to high standards as a high tech company and expect them to innovate and take risks. Right now starting a paid search model would be risky as it is an unproven market but they are probably one of the few search companies capable of taking that risk. Perhaps a small step for them would be offering a “Google Corporate” edition that strips out all the ads for companies that don’t want their employees being biased by results with advertisements.

    In the end though I can only point the finger right back at us for not demanding these rich discovery tools and being willing to whip out the checkbook for them. In today’s open-source age it does not even take a monetary investment to get involved. This is why I am so disturbed by Wikia Search’s low traction. Here we had a model where we actually had the chance to impact and get involved with the future of information discovery across the Internet. I would have expected that search professionals around the world would have lauded the project and contributed. Instead most of us ripped it apart because it did not right out of the gate live up to our expectations of commercial internet search providers.

  • 41 Carl Grimm // Mar 27, 2009 at 2:38 pm

    In the end though I can only point the finger right back at us for not demanding these rich discovery tools and being willing to whip out the checkbook for them. In today’s open-source age it does not even take a monetary investment to get involved. This is why I am so disturbed by Wikia Search’s low traction. Here we had a model where we actually had the chance to impact and get involved with the future of information discovery across the Internet. I would have expected that search professionals around the world would have lauded the project and contributed. Instead most of us ripped it apart because it did not right out of the gate live up to our expectations of commercial internet search providers.

    Let’s take a crude look at the cost of Google. Google had a 2008 Q4 revenue of 5.7 billion dollars. Google by comScore’s measure received 776 million unique visitors worldwide to all of their sites in December 2008. December is a little higher than most months so for the sake of the doubt we will assume 3 x 775 for a total of 2.32 billion unique visitors during that quarter. That is a per unique visitor profit of $2.46.

    Fade to commercial: “For just 8 cents a day you can feed a starving information worker.”

    Now of course Google now has the economies of scale that smaller companies cannot reach. Even if you assumed that a smaller provider would have to charge four times we would have to pay roughly $2.50 per week or $10 a month. Here comes the Starbucks comparison: seriously that is less than most people spend on coffee a month (or Tejava, soda, Teavana; name your poison). Don’t even get me started on cell phone bills and text messaging costs.

    Personally I would spend at least $10 for a really good information discovery tool just for my own endeavors. The return at the office would be worth a least double that.

    I think I just heard something; I hope it was the sound of the market kicking into gear.

  • 42 Carl Grimm // Mar 27, 2009 at 2:42 pm

    Correction: I said per visitor profit when that is per visitor revenue.

  • 43 Rob Gonzalez // Mar 27, 2009 at 2:52 pm

    I think another interesting facet to this conversation is how compatible the ad-supported model is to exploratory search? If I want to learn about a new topic, especially if I don’t have a specific and narrow focus for the topic, Wikipedia, not Google, tends to be my first stop. To get closer to your examples, if I’m searching for a hotel in Paris, I tend not to use Google, though I may use Google Map; instead I start with Kayak or something. For cameras, if I don’t already know the specific model (I actually think that use case favors direct text search, since you know what you’re looking for), I sortof have to start at Google, and I type things in like “Digital SLR Review” and then I get a slew of reviews, some relevant to today’s models, most old and not really relevant, but it’s a huge pain to sift through. I think those are all common examples where the Google model doesn’t work very well, ads or no ads, and I bet a huge percentage of searches are of this nature.

    So if you had something that provided a more exploratory search of what’s out there, allowing you to narrow down articles by publication date and such things, how would you actually support it with targeted ads? Could it be profitable? I’m sure there’s a way, but it’s a lot less straightforward than matching keywords to ads and monetizing the clicks. I bet that Google could become more exploratory if they wanted to; they are about the only outfit out there that has any experience computing such things at the necessary scale, but I wonder if the fear that they wouldn’t be able to monetize any of it has kept them back…

  • 44 Rob Gonzalez // Mar 27, 2009 at 2:54 pm

    Regarding ad blocking (from much earlier in the thread): I don’t believe that ad blocking, while you have to use plugins and things to get it to work, has anywhere near the mass market penetration to be effective. My guess is that those who would block the ads are those that would probably not click on them anyway, so I can’t imagine that, other than having their lawyers and lobbyists perform their due diligence, Google and others are super worried about them.

  • 45 Digvijay Lamba // Mar 27, 2009 at 3:17 pm

    To get closer to your examples, if I’m searching for a hotel in Paris, I tend not to use Google, though I may use Google Map; instead I start with Kayak or something. For cameras, if I don’t already know the specific model (I actually think that use case favors direct text search, since you know what you’re looking for), I sortof have to start at Google, and I type things in like “Digital SLR Review” and then I get a slew of reviews, some relevant to today’s models, most old and not really relevant, but it’s a huge pain to sift through. I think those are all common examples where the Google model doesn’t work very well, ads or no ads, and I bet a huge percentage of searches are of this nature.

    This is precisely the use case Kosmix is built for.

    You happen to know that Kayak is great for travel and, maybe, Amazon for product reviews. However, there are hundreds of great niche sites out there for specific domains. No one can remember all of them. For example, http://www.digitalcamera-hq.com would work for your camera use case.

    Kosmix’s goal is to bring all these little gems on one page for any query. We have thousands of partners and sources and our categorization system can understand which sites work for which categories. So we can choose the best 10-15 sites for any query you type. And we are adding new partners at a rapid pace.

    Kosmix’s goal is to be a guide to the web. Give us a topic and we want to tell you the best websites/niche search engines/databases that have information on that topic.

    You can read more about this idea at

    Kosmix Blog: The Future of Search

    Kosmix Blog: Is One Search Engine Enough

  • 46 Rob Gonzalez // Mar 27, 2009 at 4:50 pm

    I appreciate the site; I hadn’t heard of it before this thread. However, my point stands: there is a significant monetization problem for such a site since it is much more difficult for Joe Advertiser to buy ads from you. Google hits the long tail of advertisers since anybody can bid on very specific keywords or sets of keywords, and it’s really easy to understand how to do so. For exploratory search, what is the model? How does advertising scale? If not advertising, then how do you keep your servers running?

  • 47 jeremy // Mar 27, 2009 at 6:57 pm

    @Carl:

    In the end though I can only point the finger right back at us for not demanding these rich discovery tools and being willing to whip out the checkbook for them.

    I have been demanding these rich discovery tools for over 12 years, and have been signaling my willingness to pay real, out-of-pocket cash-ola for them, for at least the past 10 years.

    Personally I would spend at least $10 for a really good information discovery tool just for my own endeavors. The return at the office would be worth a least double that.

    I’ve signaled, at many times in the past, that I am willing to pay even more than that. My inquiries have been ignored; not even answered.

    @Rob Gonzalez:
    I sortof have to start at Google, and I type things in like “Digital SLR Review” and then I get a slew of reviews, some relevant to today’s models, most old and not really relevant, but it’s a huge pain to sift through. I think those are all common examples where the Google model doesn’t work very well, ads or no ads, and I bet a huge percentage of searches are of this nature.

    The Andrei Broder paper from 2002 does a query log analysis of Alta Vista, and concludes that approximately 40-45% of all web queries are of this nature. It’s no small amount. Which is one of the reasons why I keep saying Google has failed — Google supposedly does what its users want, and yet 45% of its users want this, and Google does nothing about it. Epic fail.

    Why do I think they’ve failed to offer this? Your next paragraph, Rob, hits the nail on the head:

    So if you had something that provided a more exploratory search of what’s out there, allowing you to narrow down articles by publication date and such things, how would you actually support it with targeted ads? Could it be profitable? I’m sure there’s a way, but it’s a lot less straightforward than matching keywords to ads and monetizing the clicks.

    Bingo! $25,000! Jackpot! 🙂

    Seriously, I think that’s the reason. It’s not that Google’s users don’t demand this type of search. It’s that this search is much more difficult to monetize.

    That’s why I was saying, earlier, that Google does not live up to its own motto. See comment #18, above. I reprint part of that comment here:

    Google has steadfastly refused to make any change that does not offer a benefit to the users who come to the site. But conversely, we also steadfastly refuse to make any changes that offer too much benefit to the users who come to the site. That is, we will make a change to the site if and only if it simultaneously does not harm our users, but also does not harm our advertisers. Unfortunately, this means that there are some positive changes that we could have made that would benefit the users even more, but we won’t make them because they harm the advertisers

    What I mean is that Google will not make the change to supporting exploratory search, even though it will immensely benefit its users, because it becomes very difficult to know how to advertise against exploratorily-supported search engine interactions. So because they don’t know how to monetize those interactions, they avoid developing the technology.

    Instead, they try to funnel their users into non-optimal interactions, against which they can show ads.

    There are those of us willing to pay for this service, rather than be reliant on advertisers. I’ve got shoes. I’ve got a shirt. But Google still refuses me service. It does not place my interests first.

  • 48 jeremy // Mar 27, 2009 at 7:15 pm

    The Andrei Broder paper from 2002 does a query log analysis of Alta Vista, and concludes that approximately 40-45% of all web queries are of this nature. It’s no small amount. Which is one of the reasons why I keep saying Google has failed — Google supposedly does what its users want, and yet 45% of its users want this, and Google does nothing about it.

    Sorry, let me be more accurate: Not 45% of its users. 45% of its queries. It could be 45% of its users, 100% of the time. Or it could be 100% of its users 45% of the time. Don’t know for sure.

    But statistically, there is a lower bound of at least 45% of the users, and an upper bound of 100% of the users.

  • 49 Daniel Tunkelang // Mar 28, 2009 at 1:11 am

    Daniel – you question the value and longevity of the ad-supported search model, but I’m wondering what you propose as an alternative. Let’s suppose that google.com redirected to endeca.com for the next 4 weeks. You have hundreds of millions of visitors coming to your site looking for information on the public internet – who pays for your costs?

    Ross, I see a few alternatives to the status quo:

    1) User want ads, even knowing that they represent bids paid to a web search engine for their attention. In that case, the web search engine can make them optional, and users will still want them and click on them, and thus continue the ad-supported model. The fact that no web search engine has made ads optional suggests to me that they don’t believe users want ads–only that users are too apathetic / unsophisticated / ignorant to install ad blockers.

    2) Users don’t want ads but are willing to give up privacy. In that case a web search engine may be able to make money selling analysis of user logs but still providing free, ad-free web search to users.

    3) Users don’t want ads and aren’t willing to give up privacy. In that case, search engines will have to find a way to charge users directly.

    4) Search engines mix ads into search results without telling users which results are ads. Of course, this model is thoroughly discredited. But it may return someday if that is the only way to get users to click on ads. I dislike this possibility most of all, but I don’t rule it out. Indeed, product placement seems to be increasingly prevalent in pop culture.

  • 50 Daniel Tunkelang // Mar 28, 2009 at 1:19 am

    So if you had something that provided a more exploratory search of what’s out there, allowing you to narrow down articles by publication date and such things, how would you actually support it with targeted ads? Could it be profitable? I’m sure there’s a way, but it’s a lot less straightforward than matching keywords to ads and monetizing the clicks.

    Indeed, exploratory interfaces should allow users to communicate far more precise descriptions of their information needs, which allows for much better targeted advertising. What is less clear is whether a user who appreciates an exploratory interface will be more or less receptive to ads (even if they are more effectively targeted at them) than today’s users who accept the ten blue links and still feed Google $20+B a year by clicking on ads.

  • 51 Daniel Tunkelang // Mar 28, 2009 at 1:23 am

    BTW, I have no idea how well the numbers from Broder’s 2002 paper hold up today. It does irk me that Google won’t share the most basic and seemingly harmless stats about user behavior, like what fraction of their queries return Wikipedia pages as top results.

  • 52 jeremy // Mar 28, 2009 at 11:49 am

    Indeed, exploratory interfaces should allow users to communicate far more precise descriptions of their information needs, which allows for much better targeted advertising.

    Well, I think there are two kinds of exploration: (1) Exploration in support of your info need, when you do know essentially what you want, but want as many comparative options as possible before you make your decision, or (2) exploration in support of your searching activities, when you are not yet clear what your info need is, and just need to find out more before you come to a knowledge of what the right question even is, that you want to ask.

    In case #1, I agree — ads would work. In case #2, they wouldn’t.

    Which type of exploratory search do ppl do more of? I don’t know.

  • 53 Daniel Tunkelang // Mar 28, 2009 at 12:16 pm

    Actually, I could see ads working for #2 as well, if the progressive elaboration of your information need through exploration helps an advertiser not only to identify likely paths but perhaps even to influence them. By working, I mean that an advertiser would benefit, not necessarily that this is the preferred business model for information seeking services.

  • 54 Online Advertising Fight Club | The Noisy Channel // Mar 28, 2009 at 7:17 pm

    […] of you who have been enjoying our recent debate about the future of online advertising should check out TechCrunch, which is offering a […]

  • 55 jeremy // Mar 30, 2009 at 1:32 pm

    Oh, maybe #2 could work. But would it work, at the same level of ease with which it currently works? What I mean is, right now it is dead simple for an advertiser to target his/her audience. It’s tied to very short keyword and keyword subset matching.

    Once the interaction w/ the search engine becomes more elaborate, it becomes quite unclear how you get the average Joe advertiser to be able to correctly specify when and where they want an ad to match.

    So even if the business “model” is sound in case #2, that model might end up being too complex for advertisers to actually participate in.

    Once again, I think that what is good for the users is bad for the advertisers. It’s an unavoidable conflict that Google does not have a handle on.

  • 56 Minor Changes at The Noisy Channel | The Noisy Channel // Apr 1, 2009 at 12:03 am

    […] a healthy discussion of the issues has convinced me that the ad-supported model is here to stay, I will start augmenting […]

  • 57 Brian Despain // Apr 15, 2009 at 6:03 pm

    One thing this debate has missed is that people are talking about Ad supported models of search and are assuming that Google’s CPC model is the only advertising supported model that is out there. A CPM model might work in an exploratory search model. Exploring a topic in a deep web search engine such as mednar.com means a fundamentally different user experience. Users stay far longer and use the search engine as a knowledge discovery tool. With that approach, a CPC model doesn’t make sense but a CPM model does.

  • 58 Daniel Tunkelang // Apr 15, 2009 at 8:19 pm

    That’s a fair point–to the extent that an exploratory search can be ad-supported, it may be more effective to present users with ambient display ads than PPC ads urging you to click and leave the site.

  • 59 Brian Despain // Apr 16, 2009 at 12:28 pm

    Yes because the experiences a fundamentally different. When I use google I am looking for an answer to a fairly specific question. Using something like mednar.com or kosmix.com I am more interested in exploring a topic. Further more since I know the types of information and sources used in both, I am more trusting of the results than a general web search. I am sure Kosmix users on site time can be measured in minutes not seconds like Google and that’s of real value to advertisers.

  • 60 Daniel Tunkelang // Apr 16, 2009 at 1:44 pm

    We have readers here at Kosmix–perhaps they can share some of those statistics about user behavior.

  • 61 jeremy // Apr 17, 2009 at 4:57 pm

    Users stay far longer and use the search engine as a knowledge discovery tool. With that approach, a CPC model doesn’t make sense but a CPM model does.

    But whichever model is used, CPC or CPM, the advertising that gets served has to be relevance-based. That’s Google’s fundamental premise, their core value.

    So the big question in my mind becomes: Is it possible to have display, CPM advertising that remains relevant to the user information need, in an exploratory setting?

    My intuition is that it is not that easy. No, that’s an understatement. My intuition is that it’s very, very difficult. And that’s part of the reason why I believe that Google refuses to do exploratory search.. not only is “organic” exploratory search difficult, but it is even more difficult to serve relevant ads against it, and therefore more difficult to make enough $$$ to justify that market cap.

  • 62 alexberry.me » Online Advertising Fight Club // Jan 29, 2010 at 3:47 pm

    […] of you who have been enjoying our recent debate about the future of online advertising should check out TechCrunch, which is offering a “Steel […]

Clicky Web Analytics