The WolframAlpha computational knowledge engine launch date has been moved, and the Google competitor is scheduled to go live three days ahead of schedule.
In a video posted to the blog, Stephen Wolfram discusses some of intricacies involved in getting his computational knowledge engine off the ground. One of the biggest challenges the nascent search engine faces is coordinating among all the servers it needs to be successful.
Taylor Gray, who Wolfram describes as a longtime colleague, points out that WolframAlpha will be running on thousands of servers, about 10,000 of them, handling queries as they come in once the computational knowledge engine launches.
In fact, one cluster alone consists of about 3,800 CPUs in a single data center.
"There are about 3,800 CPUs in this cluster, and they are using most of the power and air conditioning in this data center," said Gray. "If we melt down at launch it will be because of an overwhelming response."
Wolfram also revealed that the WolframAlpha computational knowledge engine has been subject to smallish scale testing, noting that the search engine had been opened up to "flicker testing" once before.
"We briefly made a small test cluster live to the world," said Wolfram in the video. "In about 10 minutes 3,000 people found their way in and generated about 18,000 queries."
Competing with Google in the search engine field is a tricky thing. Yahoo discovered that and so did Cuil, a search engine that was overwhelmed with traffic on its launch date and has since been relegated to the margins of the market.
The biggest challenge for the WolframAlpha search engine will be handling incoming traffic at launch tonight without crashing due to demand. If the engine goes down, it could be doomed to ignominy for the rest of the life of the project. But with nearly 10,000 servers prepared to go live at the flip of a switch, Wolfram hopes that his decades-long project will succeed.
WolframAlpha approaches results differently. Rather than spitting out a list of results based around a query, the computational knowledge engine will provide answers to specific questions by drawing on the accumulated knowledge of the Web.
"Fifty years ago, when computers were young, people assumed that they'd quickly be able to handle all these kinds of things," wrote Wolfram on his blog. "And that one would be able to ask a computer any factual question, and have it compute the answer.
"But it didn't work out that way. Computers have been able to do many remarkable and unexpected things. But not that."
Using that principle as his mission statement, Wolfram used the work he has done with Mathematica and NKS to create an algorithm that may not be able to provide the time a movie starts but can accurately describe the constantly changing distance between the earth and the moon at any given moment.
And that's the differentiating factor that WolframAlpha hopes to bring to the search engine market. Rather than providing thousands of answers to questions that have already been asked, the computational knowledge engine looks to provide a smaller number of results that answer specific questions with useful and relevant information.