Tuesday, June 17, 2003

Search Players Discuss Ethics of Paid Inclusion

There’s a lot of movement in the search engine industry. And sometimes, people just get together and talk. Google believes paid inclusion in search results should be clearly understood as such, while others — like LookSmart, Yahoo! and Overture — do not, saying inclusion results are still relevant, and that users are looking for commercial content. This in short is what a panel discussion revealed according to Zachary Rodgers’ “Search Players Agree on Industry Challenges” (Internetnews.com), June 17, 2003.

Helpful or not, it would be silly to say paid inclusions are just as relevant as other results, because then they would naturally appear on top and no company would spend money to artificially include them. And it’s in a search engine’s best commercial interest to spit out the most relevant results; since for the afore-mentioned reason paid inclusion will always risk that, there’s really no need to talk about ethics here.
I believe at the moment only Google is striking the right balance between clearly differentiating result page advertisements (in color, shape, and positioning) while at the same time not falling into the “banner-blindness” trap.

Search Players Discuss Et ... by Philipp Lenssen

Consulting the Web in Times of Trouble

Consulting the Web

Not all people would go online to find treatment or help to a problem, but some will. Certain problems might be urgent, for others, long term solutions would be sought. And I would think sometimes, even if it’s not time-critical, wrong information or disoriented pages can be the end of the road for the time.
If one needs help on something, chances are one won’t have a single good web page in mind to start with, so online research would be chosen. I researched the top three result pages found in Google for over two dozen queries like “suicide”, “first aid”, “stop bleeding”, “abuse”, and “gas smell”. Of the nearly 80 pages I checked, many were relevant, offering direct help and information, while others weren’t. E.g. some top ranks are taken by movies (for “house on fire” or “robbery”). Albeit only a relatively small selection of queries, there were already visible tendencies.

Positive and Negative Results

Music Sites Rank High

What stood out when investigating the sites was the large number of rock band names that took up top spots in the results. Position three for “suicide” is a rock band. Number one for “how to commit suicide” is the band by the same name. My point is not that it’s impossible someone would look for this band via Google and see the top result as relevant; no, but not finding the most relevant page for that fan is not a potential matter of life and death. And of course, the optimal site on “how to commit suicide” could be a help page which doesn’t just answer the question verbatim. (Which might be an idealistic notion considering how Google analyzes links pointing to a site. You cannot expect everyone linking online to follow your standards, even if you care to optimize the page under your control.)
Same music band preference holds true for “poison”. (The “best” results for “poison” — assuming it is an urgent case and for some reason the Web is consulted before a doctor — are for “poison help”, followed by “poison treatment”. “Drinking poison” was not giving any helpful top results, and partly returned pages that couldn’t be found.)
The rock-band AC/DC is in the number one spot for “electric shock”.

Technology Pages

Another bigger part in “irrelevant” pages (and what is relevant is clearly a subjective view) were technical references focussing on Internet technologies. “Abuse” is first and foremost about online spam, whereas “help”, “i need help” and “getting help” will lead one to programming and development sites. Specifying a more exact keyword context increases the chance of getting where one wants, but we cannot expect everyone in trouble to make the right moves intuitively. Mostly, adding the word “help” to a specific keyword drastically narrows down on helpful content.

Good Results

Relevant results that stood out positively include “heart attack”, leading to a government site with quickly accessible pages. Even the advertisement for this result on the Google result page was partly relevant. (Irrelevant to this query and others would clearly always be the news bits highlighted on top.)

Spamming

Also relatively positive, I couldn’t detect any spammers or intentionally misleading sites. There were some however that communicated their point sub-optimal. For example the number one website for “first aid” has multiple pop-ups, advertisement that loads before the rest of the page, hard-to-read colors (from dark blue on green background, to purple headlines, to a default yellow text color), and seemingly unrelated self-made cartoons. In fact, the site opened some windows that I couldn’t easily close, as it seems the browser got stuck on them.

Comparing With Other Engines

I compared some keywords with alternative search engine AllTheWeb. It’s currently not getting nearly as much visitors as Google, but should show wether or not there are huge differences in result lists. For my tests I also looked at the sponsored results, because they are so prominently displayed, hard to differentiate from normal results, and take up so much space (all the first screen for smaller resolutions, pushing actual results out of sight).
The query “suicide” shows a much more reasonable result than Google. (It’s hard to say it’s better or more justified — but for this test, I will call it more relevant). Normal results (like “Suicide: Read This First”), as well as the sponsored results (like “Self-Coached Healing of Depression”). Also, “how to commit suicide” shows support instead of punk bands. The sponsored result about how someone changed her mind about suicide is again relevant, even though — or exactly because — it does not include the phrase “how to commit suicide” and is therefore technically not completely on-target.

However, the “selected” content of sponsored results tends to back-fire on some queries. For “first aid”, instead of a help page, we are presented with first aid suppliers specifically. And on entering “poison”, the number one (sponsored) result is “Learn About Poison First Aid”, which sounds like it should really help — but if you read the description, you will see: “Pay-per-view, subscribe or purchase”.

Yahoo! as you may know picks the same result rankings as Google does, since they are using Google technology underneath. However, their sponsored results take up much more space. (On a reasonably large resolution, about 90% of the available browser area.)

AltaVista also includes a sponsored result. It’s somewhat clearer marked as such (a big red headline), but it still appears as first. What AltaVista does do is offer subject refinement via links. Entering “abuse help”, above the result we can see links to: “Animal Abuse”, “Domestic Abuse”, “Drug Abuse”, and so on.
When entering “first aid”, again, there are a lot of first aid suppliers as opposed to direct support pages. There are in whole ten supplier links with descriptions, four of them on top of the search result, and the whole page becomes cluttered. Redefining with the category-picking tool by clicking on “Emergency” brings up more sponsored results. Here, clicking on the first link that doesn’t seem to be a supplier but a First Aid Guide will lead to a shopping site — buy before you read.

Solutions and Suggestions

Webmasters

All in all, the results are mixed, and there can’t be any finger-pointing; not at Google, not at website owners.
As for webmasters occupying “hot spots” with content irrelevant to urgent cases, one could argue they should voluntarily re-focus their keyword selection. In certain cases they could also link to relevant information for those visitors accidentally stumbling on their site (even though the likeliness is rather small considering that Google clearly shows the title in the results).
On the other hand, those with relevant information could further optimize their pages, by testing keywords and keyphrases relevant to their topic.

Google’s Part

I would not favor a solution in which Google analyzes oft-used “urgent help” queries and censors certain links (which would however still be more reasonable than removing neo-Nazi content for German Google users, like it is been done currently). However; I could almost imagine rock-bands get higher rankings than help organizations simply due to the fact there’s no large linking “fan base” online for the latter. (For example, the phrase-query “suicide help” returns 0.22% of the page-count of phrase “britney spears” — not to say this necessarily is a “balanced” phrase comparison, but the weighting seems very clear.)
Here, it might be an option for Google to analyze if there’s a way to find out fair algorithms to push relevant sites.

Others

And for normal people like you and me? Well, one thing we can do is to link to those better help pages, whenever there is some relation to what you are talking about. (I do not suggest linking to a site just for the purpose of somehow increasing its ranking, because link schemes — no matter how good the intent — probably won’t really help the purpose.) And also to make sure the page you are linking to is indeed the best on the topic, and not just one from the top three results.

Consulting the Web in Tim ... by Philipp Lenssen

Google Celebrating Escher Celebrating Self-Reference

Today Google is celebrating Escher’s birthday with a special occasion logo*. Artist Maurits Cornelis Escher (one of the most-requested artists at Google) was born on June 17, 1898. The current Google logo shows the paradox of a hand sketching a hand sketching a hand... each being “at once the product, as well as the producer, of the other"**, based on Escher’s “impossible” visual construction of two Drawing Hands from 1948:

“A piece of paper is fixed to a base with drawing pins. A right hand is busy sketching a shirt-cuff upon this drawing paper. At this point its work is incomplete by a little further to the right it has already drawn a left hand emerging from a sleeve in such detail that this had has come right up out of the flat durface, and in its turn it is sketching the cuff from which the right hand is emerging, as though it were a living member.”
– Kalina Christoff, Mind, Brain and Self (Stanford University), June 20, 2000

* Past Google artist celebrations include Renaissance Michelangelo, Cubist Picasso, and Pop-ArtWarhol, among others.

** Knowledge Construction in Software Development: The Evolving Artifact Approach.

Like many of Escher’s works this one is also closely related to logic (or illogic), math (0 + 0 = 1?), programming (recursion, infinite loops, self-modifying code, and reflection), physics (perpetuum mobile), religion (if God created everything, who created God?), philosophy (meta-level of a drawing within a drawing), surrealism (it looks real but couldn’t be), the search engine world (link farms trying to create something out of nothing), linguistics (self-reference in language), psychology (self-reference and self-conscious) ...

“In [the lithograph Drawing Hands] the self-reference is direct and conceptual; the hands draw themselves much the way that consciousness considers and constructs itself, mysteriously, with both self and self-reference inseparable and coequal.”
– The Mathematical Art of M.C. Escher

... and, of course, Lego.

Google Celebrating Escher ... by Philipp Lenssen

Google Misunderstood

Misunderstanding Google URLs

Previously I researched which Google domain names are taken by spammers and in how far Google URLs might be Counter-Intuitive. Now I tested how many people link to those malformed Google URLs out of accident (I do not count Google Blogoscoped into this, because I mentioned all those URLs before):

• “googlenews.com” - 17 - real address: news.google.com or www.google.com/news/
• “googleimages.com” - 17 - real address: www.images.google.com or images.google.com
• “googleanswers.com” - 17 - real address: answers.google.com

Misunderstanding Google Copyright

Other misunderstandings of what belongs to whom include copyright issues, e.g. of pictures found in Google Images. As you may know, Google Images just find images, but they are still copyrighted to the non-Google owner. I entered the following queries (note that not all of them denote a misinterpretation of the underlying copyright issues, but some do):

• “courtesy of Google Images” - 23
• “courtesy of Google News” - 57
• “courtesy of Google Groups” - 133

There are certain complications with copyright online. Google is publishing complete websites as part of their Google Cache functionality. In general, quotes and links are considered OK — even though some sites try to prohibit “deep links"*, and some countries (like Germany) even make your responsible for linking to illegal content.

* See Site Barks About Deep Link by Farhad Manjoo (Wired), May 01, 2002.

And in general, it is not considered OK to quote complete articles (plagiarism), or link to images inline, or to “frame” another site. Because someone visiting your page might get confused as to who’s the copyright-holder. Some people suggest that in order to prevent online theft*, you should watermark your images, or use a copyright info embedded on the image (Google or not). Another consensus seems to be that Google Images does help expose images to the public, so even and specifically if you got something to sell, you might make more instead of less money from this potential copyright-misuse:

“The important thing is if you are a nobody, but have great pictures, they have to be seen by the potential buyers. If you don’t get them into a search engine as powerful as [Google], they say it’s the tops right now, then they’ll have not a good chance of competing. The reality is that people who intend to use a picture will do it. They wouldn’t pay you for it it anyway — whether you put a watermark or a copyright symbol on it or not. The people who intend to use it for commercial purposes are going to pay you. It’s not worth the risk to them to get discovered by the picture police out there on the Internet (you and me).”
– Bob Jarvis, Google Images Copyright?, September 4 2001

* Actually, whenever you are visiting a website, you already copied the images, the complete text, and all other “inline” files onto your computer hard-disk. “Theft” in this case is copying and not stealing. The difference between your computer and a webserver of course is that the webserver is public, whereas your PC (probably) is private with no outside online access.

Misunderstanding Google’s Responsibility

It has happened before that sites were asked to remove links (even Google does remove sites for certain countries). However I believe it’s a new phenomenon for a site to get attacked because it does not have certain links. Google Inc. is a private company, and yet, web hosting and advertisement company SearchKing tried to make a case against them in court attacking their “unfair” algorithms. Google argued those listing and ranking algorithms were naturally subjective in the first place. And how can anyone expect a private-run website to feature certain links to care about the financial well-doing of other private-run sites? In the end, the US Western District Court of Oklahoma concluded that “Google’s PageRanks are entitled to full constitutional protection."

After the ruling, SearchKing chief executive Bob Massa said:

“We have been able to give the Internet marketing community a much clearer view of the inner workings of Google, their systems, their reach and their own view of themselves,” he said in a statement. “SearchKing never broke a law, yet was accused, judged and executed without so much as a notice of intent. This affected thousands of innocent people without just cause.”
– Stefanie Olsen, Court dismisses Google search-fixing case (ZDNet), 2nd June 2003

Misunderstanding Google Privacy Issues

There’s a certain rumour going on that Google tracks your search queries. There have been rumours about privacy issues as long as there have been browser-cookies, those little text-files on your hard-disk which store whatever a website-owner wants to store. Now the misunderstanding is not that Google tracks something, but that it would track you. Because a cookie alone doesn’t identify you to Google. Only when you login into one of the Google services for which you created an account under your name (like Google Groups, Google Answers, or the Google Web API) can Google Inc. connect search information used on a single computer to your name. However the assumption that indeed you are connected with whatever happened on the computer is false. You might share the computer with others, e.g. family members from home, or complete strangers in a cyber cafe.

Even I track every search. Well, not me personally. It’s a default setting of most server stats to save the user IP and the referrer. So whenever someone enters a site coming from Google, the site owner will know the search terms used by analyzing the traffic logs. The referrer* will be showing the search parameters at the end. E.g.:

http://www.google.com/search?q=google+blogoscoped

(This, by the way, is also how certain pages highlight your search terms when you come from Google.)

If anything, speculations on this misuse only disillusion people of perfect privacy online, which simply doesn’t hold true as soon as you access any website, and especially when you submit forms. But even if a search string could be clearly tracked back to you as a person, then there’s nothing evil Google could ever do with the data for the time being — or how would you start to filter the approximately 6,000,000,000 monthly searches? I’m pretty sure you’d have to hire one or two billion extra cops if Google agreed to help following up on everyone entering “how to build an atomic bomb"** or anything the like. Tracking you down is much easier for any shopping site. Naturally, Google Inc.’s statement*** is not to give the information to third parties. Google, “just a soul whose intentions are good"?

* Or “referer”, as it is misspelled in the official HTT-Protocol by Tim Berners-Lee.

** Ten easy steps starting with “First, obtain about 50 pounds (110 kg) of weapons grade Plutonium at your local supplier” (Construction project, Atomic Bomb)

*** Brin is quoted to have said, “We have a privacy policy which explicitly prohibits us from ever, say, selling that information or something like that. And we also try to keep it under pretty strict lock and key.” (“The News Hour With Jim Lehrer”, 29 November 2002, via Google-Watch “Google Privacy Issues”)

Google Misunderstood by Philipp Lenssen

>> More posts

Advertisement