Enterprise Search: Search No Longer Big Enough

September 22, 2015

I read the news on LinkedIn. (Registration may be required, gentle reader.) A post by a forum moderator raised the question, “Should be expand enterprise search?” There are other signs of trouble in search land as well. The Paper.Li enterprise search curated newsletter is about Big Data, analytics, education, and—almost as an afterthought—enterprise search in the form of endlessly recycled references to mid tier consulting reports based on what are in my opinion subjective criteria.

Is the implosion of enterprise search complete? Has the shockwave of the Fast Search financial charade caught up with today’s vendors? Is the shadow of the billion dollar bust that was HP’s acquisition of Autonomy/Verity been the straw which broke the camel’s back? Was it the mid tier consulting firm’s enterprise search report which ignored the major player in open source information access? Was it the constant repositioning, faux news releases, and posturing on webinars the karate chop across the throat of search marketers?

I don’t know.

From my point of view, there are high value solutions to the challenge of providing employees with access to certain types of content. One can use the appliance approach of Maxxcat? There is Elasticsearch? The Blossom Software solution is pretty darned good. Specialist solutions are available for parts. There are even semi automated systems to help a user make sense of the noise filled streams of social media content. Think Recorded Future.

Gentle reader, starting in 2003 when I began work on the Enterprise Search Report, sponsored by, of all things, a content management specialist, there were some brand leaders. But these have fallen into disgrace, been absorbed into larger firms with little incentive to invest in research, or crashed and burned as a result of failed implementations.

What remains today are some grim facts:

  • Search is perceived by many information technology professionals as a problem. Enterprise search implementations are often doomed from the git go because few want to hook their careers to projects which have for decades failed to keep users happy and been unable to provide useful results without constant infusions of money, computing cycles, and whiz kids.
  • Open source solutions are available, and they are pretty good. Large companies have the time, staff, resources, and incentive to get away from the proprietary solutions which limit what the licensee can do with the system.
  • Search is an inclusion in the most advanced systems. Consider Recorded Future, Diffeo, or any other cyber centric, next generation system. System is available, but these systems solve specific problems. Search is sort of an apple pie, mother, and love type solution. These generalizations are tough to apply in a business like manner in organizations struggling to pay their bills. Most organizations just use what’s available? Even AutoCAD includes a search function. Oracle, bless its proprietary heart, provides a database licensee with a good enough solution. For those wanting a more robust solution, the Secure Enterprise Search system is available without charge. Yikes.

In my own experience, the sins of the earliest enterprise search vendors like Fulcrum Technologies and Verity have bulldozed a highway built on quicksand. Today’s vendors talk about search in terms of buzzwords like these:

  • Customer support. The idea is a variation of ClearForest’s pitch that one can find answers to customer issues by indexing text.
  • Big Data. I am confident that when I look for information in a Big Data set, I want to use search as a secondary tool. Enterprise search vendors offer analytic routines as add ons or as spin on counting terms which have been extracted.
  • Taxonomy. I love this concept. A company needs to index its content. Nothing improves search, which has not been improved too much in the last 50 years, like machine indexing. Just don’t pin down the vendor on the amount of human intervention that is required to keep the automated system on track.
  • Natural language processing, semantics, and artificial intelligence. The idea is that a search system with smart software can figur4e out what a human generated document means and make it  available to a user easily or, in some cases, BEFORE the user knows she needs access to the information in that document.

There are three problems which vendors and their customers have to wrestle into submission.

First, vendor and customer have to agree on exactly what the information access system is supposed to do. In my experience, this is an important step which is usually given modest, if any, attention. The reason is that instead of narrowing the focus to a specific problem, the problem gets defined in ever widening circles of functionality. The result is cost overruns and disaffected users.

Second, the vendors’ marketing argues that certain functions and benefits are a consequence of installing their software. The flaw is that marketing is easy; implementing a search system which the customer can afford to maintain is very hard. Add to this disconnect the characteristic of some vendors to sell software which is half baked, or, in some cases, not even completed. A certain vendor was kicked off a government procurement list for getting caught with software that did not work.

Third, the customers know that finding information is important. Most enterprise search vendors cannot provide access to the type of content which is growing rapidly and gaining importance with each passing day. I am talking about indexing audio, video, social media generated by employees and contractors, and digital images. The focus has been for a half century on text. That does not work particularly well if one does not select a solution from a handful of vendors with solutions that actually work. Need I repeat Blossom, Elastic, and Maxxcat?

What about today’s flagship vendors? If one embraces the analysis of the mid tier consulting firms, the solutions are ones that are proprietary and have some profile and money due to the ministrations of addled venture capital players looking for the next Google.

There are solutions. Until the LinkedIn pool of job hunters and consultants comes to grip with software that works in a reliable manner, it is unlikely that the enterprise search discussion on LinkedIn will rise above thinly veiled marketing.

Search, gentle reader, is important. There are solutions which work. The problem is that in today’s go go world, those with a veneer of knowledge and expertise are guided by individuals who may be failed webmasters, unemployed journalists, English majors, and self appointed experts.

I have no solution to the crisis in enterprise search. Google muffed the bunny. Microsoft has its Powerset and Fast Search technologies. IBM offers Watson.

Maybe these solutions will work for you. They won’t work for me. Search experts, crisis time. My vantage point is from rural Kentucky. The experts in Manhattan and San Francisco have a much better view. What they see, however, is quite different from what I observe. Just make search bigger. The problems will just fade away, right? Grass is easy to grow in scorched earth, correct?

Stephen E Arnold, September 22, 2015

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta