Wolfram Alpha: Google in No Danger

April 26, 2009

ReadWriteWeb.com’s review of the Wolfram Alpha search system is here. The name Wolfram Alpha includes a pipe symbol, which I am not going to include in this write up. Indexing systems have enough trouble with words. Including a vertical bar is one of those marketing things that annoy me. Booz, Allen & Hamilton in the 1970s replaced the comma between “Booz” and “Allen” with a dot. What a headache. No vertical bar for me, thanks.

The headline on the write up was “Our First Impressions.” In sales and search, first impressions matter. The problem is that each search engine is different, and the methods required to get the results a user wants takes time, experimentation, and often industrial strength testing.

ReadWriteWeb.com’s Frederic Lardinois participated in a Web demo and concluded:

It will definitely not be a Google killer.

He continued:

Alpha is built around a vast repository of curated data from public and licensed sources. Alpha then organizes and computes this knowledge with the help of sophisticated Natural Language Processing algorithms. Users can ask Alpha any kind of question…

Alpha taps into a buzzword that like a summer tornado has been gathering momentum before it hits the trailer park of content that makes up much of the information available on the public Internet. To duck into the storm cellar, Alpha focuses on sources that are “curated”. My impression which is not even based on a demo is that Alpha is in the “deep Web” business. The idea is that there are some useful sources which may be tough to index with a general purpose indexing system like that used by Microsoft Live.com or Yahoo.

When results appear, the system attempts to answer the implicit or explicit question. AskJeeves.com focused on this angle in the early 1990s, but quickly ran aground due to editorial costs and the reluctance users had to ask questions the template system could answer; for example, “What’s the temperature in Chicago?” worked. Questions like “What is the IBM patent for RDBMS?” did not work.

ReadWriteWeb.com pointed out that there will be a free and a for fee version, alerts, and a way to “embed” Alpha into other applications. According to ReadWriteWeb.com, the demo “mostly focused on math and engineering data, so we’ll still have to wait and see how Alpha copes with questions about historical events, for example.”

Let me make several observations:

  • There continues to be a hunger for a system that answers questions. Users don’t ask questions, but those in the research and investment business assume or know that users * should * ask questions. The ArnoldIT.com worked on a couple of question answering systems in the past, and we learned first hand that users want the system to make life easy. The idea search system just displays information the user needs to know at a particular point in time. Today the predictive mobile search systems that hook into a context clue like a geographical position seem to be closer than a search box.
  • Questions pose big problems in ambiguity. There are many ways to serve this fish. Google has its PageRank core wrapped in add on methods to give you a rock star when you type “spears” or “idol”, not Macedonia weapons. The proof of the disambiguation will emerge not from a demo or from impressions but from subjective and objective tests. So the fish remain in the stream at this time.
  • The hope for a Google killer is now a catchphrase. The problem with killing Google is that it has morphed into more than search. The Google platform is going to be tough to break up because it is polymorphic, a characteristic I I discuss along with fluxion in my new study Google: The Digital Gutenberg here. Google has some interesting data management tools that can deliver answers as well.

To wrap up, the ReadWriteWeb.com story is good. The Alpha system seems interesting. The desire to have a kinder and gentler Web search engine is intense. The challenge will be delivering users who understand an answer and who have the types of questions a system can answer without sucking too much money from its investors until the cash begins to flow.

In the meantime, “Cuil” is not longer the word. The word will be “Alpha” for the pundits until the next Google killer walks out of the computer lab into the world of the average Web user.

Stephen Arnold, April 26, 2009


One Response to “Wolfram Alpha: Google in No Danger”

  1. Mark Johnson on April 26th, 2009 10:19 pm

    Agreed entirely on your assessment. I’ve now worked for two companies, Kosmix & Powerset, that have been branded as “Google killers.” Sometimes this comparison is invited by the companies themselves (Cuill, Wolfram|Alpha, and even Powerset) and other times it’s the press pushing the issues (Kosmix is smart about not wanting to be in this category). My personal opinion is that it’s extremely unlikely that a “Google killer” is going to come out of a small amount of venture capital, much as it’s unlikely that a small company is going to be able to produce a competitive chipset or operating system. I wish companies would just focus on what they do right instead of trying to promise the stars. Nonetheles, I’m excited to see what Wolfram has cooked up this time!

  • Archives

  • Recent Posts

  • Meta