Learn how Wolfram|Alpha functions and the principles behind its unique search engine capabilities.

On March 5, 2009, Stephen Wolfram shared on his company’s blog about an upcoming product—an engine designed to answer queries in natural language, soon to be released by Wolfram Research.

Google dominates the search market in the United States, holding 64.2% of the market share in March 2009. Yahoo comes in second with 15.8%, and Windows Live holds 10.3% [source: Nielsen Online]. Despite this, Wolfram's announcement made waves in tech circles. If giants like Yahoo and Microsoft haven’t dethroned Google, why should we pay attention to Wolfram Research and its new website?

People are often drawn to stories where the underdog triumphs over a seemingly invincible opponent. In this case, however, David isn't entirely defenseless—he has a powerful slingshot. Back in 1988, Wolfram Research introduced Mathematica, a groundbreaking software that revolutionized data manipulation. When creating Wolfram|Alpha, the company applied similar principles to handle data from users' questions and present answers in fresh and innovative ways. Thanks to the trust associated with Wolfram Research, Stephen Wolfram's announcement caught everyone's attention.

What exactly is Wolfram|Alpha? What does it do, and how does it stack up against Google?

Computational Knowledge Engine

Wolfram|Alpha, launched on May 15, 2009, faced early criticism from users who expected it to function like a search engine. These users were disappointed when it couldn't provide answers to basic queries. However, unlike Google and similar search engines, Alpha isn't designed to index global information. Instead, it serves as a computational knowledge engine, as defined by Wolfram Research.

Traditional search engines gather websites from across the Web and catalog them in their directories. The algorithms behind these search engines rank the pages based on various factors. One such factor is how often users click on a particular result from the search engine results page (SERP). Another factor is the number of external sites linking to a particular website. If you build a public website, it is likely to be included in search engine results automatically, as long as it has links from other sites.

Unlike search engines, Wolfram|Alpha does not scour the Web for information. Instead, it uses licensed databases and relies on content entered, tagged, and organized by Wolfram Research staff. At the time of its launch, Alpha's servers contained more than 10 trillion data entries. Wolfram Research employees verify and ensure the accuracy of all information before it becomes part of the Wolfram|Alpha database.

To retrieve information, users simply input a query into the Wolfram|Alpha search box and press the equals sign, signaling that mathematical computations are occurring in the background. Alpha then searches its databases for relevant data to answer the query.

The results can be straightforward: entering the name of a city provides details such as the time, weather, population, and elevation. If multiple cities are entered, Alpha presents a comparison table. Entering an animal's name will give you details like average size, alternate names, scientific classification, and taxonomy. If multiple animals are queried, Alpha provides a comparison table, showing how they are related in the taxonomic hierarchy.

Alpha excels at solving scientific and mathematical problems. You can ask the computational knowledge engine to solve a calculus problem, and not only will it provide the solution, but it will also walk you through the necessary steps to reach that answer.

Keep in mind, Wolfram|Alpha is not a search engine. So, if you're looking for celebrity images, the lowest airfare to London, or articles on a particular subject, you’ll need to use a search engine. Alpha is designed to provide factual answers to your questions, accompanied by visual interpretations of those answers.

What's behind Wolfram|Alpha?

At the age of 15, Stephen Wolfram published his first scientific paper. By 20, he earned a PhD in theoretical physics from the California Institute of Technology. In 1981, Wolfram was awarded a MacArthur Prize Fellowship, a pivotal moment when he began exploring the complexity of natural systems.

Stephen Wolfram started using computers in 1973. In 1986, his company, Wolfram Research, released the first version of Mathematica, a software widely respected across industries such as engineering, science, and finance for its advanced computation and modeling capabilities. The algorithms within Mathematica are also utilized to process and display search results in Wolfram|Alpha.

In 2002, Stephen Wolfram authored "A New Kind of Science," a book where he outlines his belief that simple rules can unlock the understanding of complex problems. In a blog post from March 2009, Wolfram explained that the insights from his book, combined with the computational power of Mathematica, led him to believe that creating Alpha was achievable.

One of Wolfram's primary objectives was to enable users to "ask a computer any factual question and have it compute the answer" [source: Wolfram]. However, natural language processing is complex. Wolfram Research revealed that the website uses algorithms and analytical software to detect data patterns, helping it interpret common shortcuts in user queries and identify the information being sought.

For instance, if you type "GEC" into Alpha, it will assume you're referring to the General Electric Corporation. If you add commas between the letters, Alpha displays musical notes and shows where they appear on a keyboard. By clicking the "Play Notes" link, you can hear the progression. For Americans, it's easy to guess that GE is the parent company of a major TV network. Wolfram|Alpha infers what you're searching for based on the way you phrase your query.

According to Wolfram Research, Alpha is built on four main principles: data, dynamic computation, natural language understanding, and computational aesthetics. While the data is managed by Wolfram Research staff, the rest of the process is handled on the backend with the help of Mathematica.

To handle such massive calculations, Wolfram Research employs impressive computing power. They utilize two supercomputers, built with customized Dell hardware by R Systems. Named R Smarr, these machines rank as the 66th-fastest supercomputer in the world (as of this writing). Equipped with 4,608 processor cores, R Smarr can perform 39.6 trillion operations per second. Inside each supercomputer are 576 quad-core Intel 'Harpertown' Xeon chips and 65,536 GB of random-access memory (RAM) [source: Shankland].

Wolfram|Alpha vs. Google

With so many people eager to compare the search engine giant Google with the up-and-comer, why not take a closer look at both and see how they measure up?

According to Wolfram|Alpha (using Alexa.com as a source), as of this writing, google.com receives 3.8 billion page views daily, while Alpha garners 13 million. In comparison, Google holds the top spot on the Web, while Alpha ranks 568th.

However, this doesn't tell the full story. Google generates revenue primarily through advertising, earning $21 billion in 2008, with 99 percent of that from ads [source: Buley]. Wolfram Research also uses advertising to support its site, with a contextual "featured sponsor" box appearing beside certain search results.

Wolfram|Alpha may need to explore additional revenue streams. Around 250 people work behind the scenes at the site [source: Shankland and Needleman]. Manually curating data is more expensive and time-consuming than using web crawlers to index sites. The database that Alpha launched with took three to four years to compile [source: Beaumont].

Wolfram Research might also offer customized subscriptions to organizations wishing to compare their proprietary data with the Alpha database. Additionally, there's an API available, allowing developers to integrate Wolfram|Alpha's functionality into their own websites [source: Shankland and Needleman].

Before Wolfram|Alpha's official launch, Sergey Brin, Google co-founder and President of Technology, was given a demonstration of the technology by Stephen Wolfram himself [source: Levy]. This could indicate a potential partnership. On the other hand, some see Google's introduction of Google Squared—a feature that enables users to compare data in ways similar to Alpha—as an attempt to overpower the newcomer. However, Google Squared relies on publicly available data, unlike Alpha's use of exclusive proprietary databases [source: Talbot].

As Stephen Wolfram mentioned in his March 2009 blog post, Alpha is a vast project that will never truly be 'finished.' Whether it will manage to gain a place in the hearts of Internet users or if its technology will become the backbone of search efforts on other platforms remains to be seen.

For more information about Internet search technologies and related topics, continue on to the next page.