IBM: From Mainframes to SEO

February 21, 2010

IBM’s alleged mastery of SEO baffles me. I remember hearing a talk several years ago from another IBM professional . I think the person’s name was Morano, Morone or Morrano (not Murano, that’s the glass place near Venice). I blanked out of the talk because IBM makes life really tough for me to locate information on its Web site. An outfit with an almost unusable search system is not going to have much credibility lecturing me about getting indexed in Google. I received via my trusty Overflight service a snippet pointing me to Writing for Digital. This blog ran an article I found darned remarkable. The story, which I urge you to read, is “Case Study: 2 Kinds of Organic Search Competition.” I am not “into” search engine optimization. I am into creating what I think is useful content for myself. If others read what I develop, okay with me. If others don’t read my information, okay with me. I am in the minority, but I think more folks should create original information and skip the SEO work that fascinates so many experts.

Several comments about this write up:

First, it uses the phrase “link juice” as a tag. I am not sure how many experts, even the SEO experts, use the phrase “link juice” when searching. In the database business, we would use unusual terms to keep tabs on wily vendors. Humans, nope?

Second, presumably “link juice” is in line with the author’s recommendations for high value SEO. This passage caught my attention:

One of the principles of our book is to do keyword research before you begin to even concept, let alone produce a Web page. In the old days, this didn’t happen very often on marketing pages. Traditionally, the messaging for a campaign was determined and the framework for the campaign’s Web copy was written before the search experts were brought in to choose the best keywords for it. That made it difficult to attain tight relevance between the copy and the keywords—leading to poor organic search performance.

Yep, “link juice” matches this info.

Third, how much work is required to get a $100 billion outfit’s boss indexed. Obviously a whole lot. Consider this passage:

Every speech by Sam Palmisano receives a lot of media attention, with good reason. So, the landing page was linked to by such media outlets as BusinessWeek, and others. It had blog mentions galore, including ReadWriteWeb, and others. Within two weeks of the speech, the landing page had more external link equity than all but the most central pages in, which have link equity mostly by virtue of the pages in that link to them, not as much by external links. After a few weeks, the speech landing page had more external link equity than all but a handful of pages in

In my deep experience as an addled goose, I think this is more craziness than this addled goose can tolerate. You may be different. That’s what makes goose races so exciting.

In short, I encourage you to follow in IBM’s footsteps is you [a] have a $100 billion revenue stream at your back, [b] Provide an almost unusable Web site to the hapless folks looking for documentation for an IBM device requiring a FRU, and [c] are trying to cook up a reputation as a guru in a field that is filled with land mines, potholes, and engineers who prefer clean code and original content.

Wowza! Why not make some changes to and provide substantive content?

Stephen E Arnold, February 21, 2010

No one paid me to write this. I suppose I should report non payment to the GSA. IBM just landed a big contract to fix up the GSA’s computer systems. I will probably send my inputs to an IBM system which will certainly  work like most IBM search systems.


3 Responses to “IBM: From Mainframes to SEO”

  1. James Mathewson on February 22nd, 2010 12:17 am

    First of all, the gentelman’s name is Mike Moran, and he’s one of the five leading experts in search in the world. He deserves at least to have his name spelled right, which you could have checked with a simple Google search.

    I agree that IBM’s current internal search system stinks. That’s why I have worked for two years to get a total overhaul of the system. The new system for the masthead will be available in May. The new system for the whole collection of 4 million Web pages, plus hundreds of thousands of other kinds of information, will be probably another year of development before it’s ready for prime time.

    On the masthead, we are ripping out our current engine based on Microsoft Fast (which we had long before Microsoft bought the company) and replacing it with Google Site Search. Not the Google appliance, but a search expeirence that should at least equal Google’s external experience, with some IBM enhancements. Our internal tests indicate a complete reversal in the quality and relevance of results from the Fast implementation.

    I can’t say much about the solution we’re developing for the rest of the environement, because it’s proprietary. But I can say that it is an open-source solutions based on technology developed by IBM Research that should lead the industry in search in the next decade. Our initial tests indicate that it outperforms Google for high-context searches, which is the benchmark.

    Finally, I’d like to challenge you to develop a search system for an organization that publishes as much information as IBM. IBM is the largest and most diverse publisher of original Web content on the planet. If our new system works for IBM, it should work for the rest of the workd. That is our sincerest hope.

  2. Stephen E. Arnold on February 22nd, 2010 8:50 am

    James Mathewson,

    Thanks for writing.

    I noted: “I’d like to challenge you to develop a search system for an organization that publishes as much information as IBM. IBM is the largest and most diverse publisher of original Web content on the planet. If our new system works for IBM, it should work for the rest of the workd. That is our sincerest hope.”

    I think “workd” is “world”, correct?

    The addled goose does appreciate your challenge, but I have already had some modest experience with IBM’s own search expertise. Your firm had me and my team work with several of your colleagues on the ITRC mainframe search system in the 1981-1982 time period. That was decades years ago. Nevertheless, we have some perspective on STAIRS III, Web Fountain, iPhrase, FileNet search, and the work of former IBMers like Dr. Ramanathan Guha, formerly at IBM Almaden and now at Google. Then, for IBM we were able to hook BRS Search into our document delivery service. My recollection is that an engineer at IBM could not find documents prior to our work. IBM remained one of our largest customers until the sale of the online unit of the Courier Journal & Louisville Times Co. to Gannett and Bell+Howell. My recollection is that the home grown IBM document retrieval system in the IBM ITRCs (global library system) made IBM users somewhat unhappy. My recollection is that we fixed the problem.

    After that work, I got a call from the pre-break up AT&T Bell Labs. Someone at IBM sent the IBM MVS TSO operations manager to me. The IBM software providing search and reporting services made some of the Bell folks have upset tummies. Our fix was to “hook” Information Dimensions BASIS into the MVS system and chop out IBM’s subsystem. Our solution worked and the AT&T grouches went back into their cubicles.

    As for spelling names, the addled goose is a humorous fowl. I was trying to dance around the “moran” – “moron” pun. I guess I was too subtle. I won’t make that error again. Thank you for reminding me to be explicit in my personal Web log.

    I am delighted that you are doing the “rip and replace” approach to fixing search. In the monograph Successful Enterprise Search Management, I wrote with co-author Martin White, we pointed out that often companies with ** deep pockets ** can afford the rip-and-replace approach to search remediation. Your comment adds a useful data point to our analysis. If we do another edition of that monograph, I will include the IBM example as a mini-case example.

    I appreciate your pointing out in a public Web log that your incumbent system does not work. An IBM legal eagle might find that statement worthy of a closer look. If I were back in the managing people game, I would be cautious about certain vendors’ shortcomings in the technology performance area. But what do I know as I sit next to the mine run off pond?

    The other interesting factoid is that IBM will not using the Lucene search solution, which is, I believe, in the IBM OmniFind product. Why not put Google’s technology in that system? Are customers going to have to deal with mash up search systems with a little bit of iPhrase, a dollop of Lucene, and a dash of WebFountain? What’s good for the goose is good for the gander! Maybe customers are not “gander”?

    Stephen E Arnold, February 22, 2010

  3. Affordable Search Engine Marketing Services on April 25th, 2010 10:00 am

    Affordable Search Engine Marketing Services…

    1. ……

  • Archives

  • Recent Posts

  • Meta