The results of my first taste of Wolfram Alpha prior to its launching support my view in a previous post: It is relatively good in mathematics, Physics and other sciences. However, even in these fields it is far less good than expected. Some people who glorify it were deeply disappointed.
I tried in vain to find answers to very basic non-scientific questions.
The games of Chess and the game of Bridge are unknown to its Answers Engine.
It is unfair to judge a product even prior to its launch; therefore I am not going to analyze its ability to answer different type of questions. I will wait about half a year before analyzing it deeply.
A new killer in town: Wolfram Alpha is the Google Killer.
Is it really a Google Killer as dubbed by some people? Although, my knowledge of Wolfram Alpha is less than limited, I am sure it is not.
Some arguments supporting my view:
1. Wolfram Alpha is not available yet for public use even as a Beta.
People impressions are based on previous achievements of StephenWolfram, Demos, an empty Web site and a blog.
As far as marketing is concerned, it is a huge success: so many references in the Web before it was released.
2. Is it a Search Engine? A Knowledge Base? An encyclopedia?
If it is not a search engine it is not competing directly with Google.
3. Frequently IT application Killers or Infrastructure Killers were doomed to death before their competitors they tried to kill.
For examples, IBM's 9370 computers described in the nineties as VAX killers fade away about 10 or 15 years before VAX computers end of life.
An article I read many years ago was titles: RDBMS death. Object Oriented Databases (OODBMS) should replace them according to the article. Today some people may argue that OODBMS are not dead, but RDBMS are still alive and kicking.
I also read some articles about C# as Java Killer. Java is still with us and will be around for many years.
4. Only Alpha's first part will be introduced.
I am quoting Wolfram Alpha blog: "And—like Mathematica, or NKS—the project will never be finished. But I’m happy to say that we’ve almost reached the point where we feel we can expose the first part of it."
-->
5. It aims at addressing theoretical scientific problems e.g. Natural Language understanding.
Translating a scientific breakthrough (and my scientific background is too limited for judging if it is a breakthrough) to commercial products is always a challenge.
-->
-->
What is Wolfram Alpha?
-->
My understanding is based upon Wolfram's blog, demo screen shots, a YouTube video including demo presentation by Stephen Wolfram and few other Web items.
It is a kind of search engine, but the search is based upon questions instead of a search by keywords. The answer to a question summarizes data on the topic of that question.
For example if you ask: What is the GDP (Gross Domestic Product) of France? The answer will include a number representing that value and some histograms and related variables values. It is possible to drill down to more specific questions and get more detailed data.
The knowledge supplied resembles Wikipedia more than Google search engine.
The following factors distinguish between Wolfram Alpha and Wikipedia: -->
The Alpha's knowledge base could be deeper and more scientific than encyclopedic data.
No open community is mentioned as participating in Wolfram Alpha
Wikipedia's search mechanism is based on keywords
No explicit digging or drill down mechanism is inherent in Wikipedia. However, in some articles in Wikipedia the text includes references and hyperlink to other articles describing embedded topics. For example an article on SOA may refer to an article on SOA Contract or on SOA Government.
-->
As already mentioned, the questions are asked in natural language (English in this specific example).
The answers are based upon the vast knowledge available in the Web.
Wolphram Alpha utilize methods and algorithms of a previous project by the same company: Mathematica.
Mathematica is used for mathematical computation, modeling, simulation, visualization, development, documentation, and deployment.
The approach is based on calculation algorithms and is different from the Semantic Web approach.
My Take
Wolfram Alpha could be an innovative and successful product in the future, but it will not be a Google killer or even Wikipedia killer.
It is true that rarely a new product may be a killer of an older one, but special conditions are required for supporting this process. I do not notice such special conditions in Wolfram's case.
. I can think of two well known examples:
1. Microsoft's Internet Explorer as Netscape's browser killer. The special conditions in this case were a dominant and stronger vendor (Microsoft) and a wrong approach of the market leader (Netscape). Notice that after Microsoft's victory, its market share is cannibalized by newer solutions (Open Source's FireFox and Google's Chrome).
2. Google's Search engine was a killer of some of preceding products. Google's first success
upon better innovative search algorithm (Page Rank), Larry Page & was based Sergey Brin's decisiveness, and relationships which enabled sponsorship of adequate Venture Capitals, as well as hiring of a an experienced
CEO (Eric Schmidt).
Understanding Natural Language is quite a big challenge
Wolfram blog refers to Alpha as something that "almost gets us to what people thought computers would be able to do 50 years ago!"
Looking back only 25 years, I learned a little bit about academic disputes on computers abilities to understand natural human languages.
It is relatively easy to understand the syntactic layer but understanding the Semantics is more difficult. Simple example (I do not remember who the originator of that example is) cans illustrate the difficulty.
The following two syntactically identical sentences have different semantics:
The glass fell on the table and it was broken.
The rock fell on the table and it was broken.
The first sentence tells us that the first object (a glass) was broken.
The second sentence tells us that the last object (a table) was broken.
I do not know if Semantic Web is a good enough solution to understanding the semantics of natural languages, but it seems reasonable that a solution for that problem is required.
Wolfram Alpha may provide good understanding of mathematical and physical (and probably other scientific) questions but probably less good understanding of natural language based questions in other fields.
Questions in scientific fields are usually more accurate and formalized than questions in more fuzzy topics. Therefore it is easier to understand and interpret questions in these fields.
A lot of data is available in the Web but not all data was created equal. Some of it is knowledge or valuable information and some of it is useless.
The challenge is to distinguish between reliable information and non-reliable data. I do not know yet, how Wolfram Alpha is going to address it. In any case, it seems like its ability to distinguish between different kinds of data will be better in scientific fields than in other fields.
Google is not a Search engine only company therefore a fierce competition of a new Search Engine not necessarily kills it.
For example, an improved search engine without proper advertising mechanisms and advertising mind share may require cooperation with Google in order to use its advertisement expertise and tools.
A breakthrough in Search Engines algorithms could hurt Google even if it will not kill it.
A breakthrough will probably be based upon some kind of semantic search instead of keyword search. Wolfram Alpha could be an example of product including more semantic search and search results capabilities. Expect for other semantic search tools in the future.
In order to survive (see my post Vnedors Survival: Will Google survive until 2018?), Google should continue researching and inventing new searching algorithms, as well as coping with new algorithms used by competitors.
A final concluding remark
Wolfram Alpha may be a promising inventive product, but looking back to 1957, it could also be another General problem Solver, i.e. a pretentious general purpose effort, capable to answer formalized mathematical questions, but far away from answering real world problems.
This post is another post in the Vendors Survival series following posts on Microsoft, Google, HP and Sun.
I read a post by Jason Hiner the Editor in Chief of TechRepublic in his Zdnet blog. The post presents 7 possible vendors' acquisitions and merges during 2009. EMC is a player in hypothesis number 1:Dell and EMC merge and in hypothesis number 2: Cisco buys Vmware.
I do not think these two hypotheses will be realized. I do think that EMC management team will not make these two mistakes in 2009, if the company would like to survive until 2018.
Why EMC should not sell Vmware?
Jason Hiner wrote "EMC, the world’s leading storage vendor, is flush with cash and has the opportunity to make a big move". If a company has a big amount of cash, it should have a good reason for selling a company it owns.
I do not find any reasonable Business Case for selling Vmware.
It is true that Vmware is no longer the only leader in Virtualization functionality, but it is still the market leader. The main competitors are Microsoft and Citrix.
Microsoft is a competitor of Windows only Virtualization, so it is not a competitor as far as heterogeneous platforms Virtualization is concerned and it is not a competitor in UNIX or/and Linux Virtualization.
It should be remembered that Virtualization is a growing market, so the probability of higher Vmware revenues is high.
It should also be remembered that EMC strategy was based on changing from a storage hardware only vendor to a vendor with software portfolio. As a result of that strategy it acquired companies such as Documentum, RSA, Legato and Vmware.
My conclusion is that EMC is not going to kill the golden egged chicken Vmware and will not sell it in 2009 to Cisco or any other company.
Why EMC should not buy Dell?
Jason Hiner wrote about merging of the two companies and compares it to HP-Compaq merge. However, HP headed by Carly Fiorina acquired Compaq and did not merge with Compaq; same as if the scenario of merging between EMC and Dell will take place it will not be a merge but acquisition of financially weaker company (Dell) by the company having much more cash (EMC).
It is true that many years ago EMC acquired Data General which produced servers for proprietary Operating System (AOS/VS) as well as UNIX servers, but EMC acquired it only because of the midrange storage line and did not compete in the server market utilizing DG UNIX servers.
Acquiring Dell is a different story. It does not have a leading storage product line for SMBs and it is bigger than DG, therefore its acquisition will be more complex, will require allocation of more resources and time for completing a merge.
Will it enter into new crowded markets (Servers and PCs) during a Recession? I doubt EMC will try to enter to these markets which are not growing fast in such an unstable period.
As a company which is focused on the High-End market, the PCs division could be a difficult challenge. Some of the potential clients will be little Computers shops and many of the customers will be home users.
Should it kill Dell's PC division after an acquisition? Probably not, Dell is a leading vendor in that market.
Will it buy Dell and look afterwards for a Chinese or Indian company which will buy the PC business line? There is no good answer to these questions.
EMC could continue its partnership with Dell but it would not and should not buy Dell.
What should EMC do in order to survive?
The key for EMC's survival until 2018 is keeping its position as a leader in the Storage market. However, the storage market is changing and storage vendors should innovate in order to keep their current position or even improve it.
Storage solutions should be adapted to a more distributed environment with a varied data types and formats including less structured data than typical data stored in traditional Relational Databases. Sometimes the data is distributed in a Grid and in many cases it there is a Web based data.
EMC could use its cash for acquiring innovative storage companies (HIV acquisition by IBM is an example of such an acquisition). It also may acquire software companies, especially software products for information management.
Are the days of the High End market disks over?
The answer is definitely not. Large enterprises storage is growing in high rates (about 20% a year) and some of midsize enterprises growth rate is even higher.
Cloud Computing market growth rate is very high. Clouds will utilize High End disks.
It is obvious that EMC will aim at the Storage for Clouds market segment. For that purpose its traditional storage devices should be adapted to the characteristics of that market. However, the new division's name includes also the term services.
My hypothesis is that the company may also consider providing Cloud Computing services same as Amazon, Google, IBM, Microsoft and other companies are doing.
In that case one of the smaller Cloud Computing providers could be a target acquisition for EMC.
Virtualization is a must in a Cloud (External Cloud as well as Internal Cloud).
Vmware could serve as a Virtualization software company for EMC, creating new and extended Virtualization solutions for EMC's Cloud Computing services, as well as for selling these future products to other Cloud Computing providers.
The scenario cited above is additional reason why not to sell Vmware.
Probably, each Cloud Computing provider will provide specialized services based on its previous expertise. Microsoft's and Google's Cloud services are good examples of providing Cloud services based on former expertise.
What kind of specialized Cloud services would EMC provide, if any?
The following answer is a guess or a speculation. Probably my guess as a SOA expert and not a Storage expert (although I have some experience in evaluating, selecting and properly utilizing Storage in a Data center) is not better than other people's guesses. Anyway, I think of the following types of Cloud Computing services EMC may provide:
Infrastructure as a Service (IaaS) – storing large amounts of data in the Cloud Computing Data Center. Backups of data located in the Data Center could also be stored in the CloudComputingDataCenter.
Providing data synchronization services between DataCenter applications and applications data in the Cloud.
Data Synchronization with a Cloud is an emerging challenge today, when External Clouds usage for Core applications is rare in large enterprises. The reason for not using the Cloud for Core applications is there interconnections with other core applications, which are available only in the DataCenter. It will be a major challenge, when some Core applications of large enterprises will run in the DataCenter and others in a Cloud (If different applications will be executed in different clouds the challenge will be magnified).
Data Synchronization may be achieved by improved techniques of near parallel write operations in the DataCenter and the Cloud executed transparently by the Storage Servers or Controllers.
It should be noted that a capability of that kind for a reasonable cost could provide a solution to the synchronization challenge.
Content storing and Management
This service could be based on Documentum software which is included in EMC's portfolio. Its RSA products line could used for protecting the content from unauthorized access.
