I felt pretty lucky to get an early preview of Wolfram Alpha, but real insiders like Rudy Rucker get personal demos from Wolfram himself. Rucker not only published a post about his conversation but also posted a “slightly condensed” hour-long podcast of the interview.
I listened to the whole hour. My biggest surprise is at how much Wolfram emphasizes the natural language interface, which I’d thought from my own preview was less of a focus–and even the system’s weakest link. In fact, when Rucker asks if there will be a manual offering advanced users a list of expressions akin to those in Mathematica, Wolfram says no, that users are lazy, ontologies never line up, and that the system will figure out what the user means. Of course, I now question my assumption that Wolfram Alpha’s real goal or potential is to act as a service to use in other applications.
At the same, Wolfram says that he envisions “a new field of knowledge-based computing. Imagine a spreadsheet that can pull in knowledge about the entries.” I heard that and then expected him to explain a vision around an API, but he continued to explain that the interface from a spreadsheet woul…
I felt pretty lucky to get an early preview of Wolfram Alpha, but real insiders like Rudy Rucker get personal demos from Wolfram himself. Rucker not only published a post about his conversation but also posted a “slightly condensed” hour-long podcast of the interview.
I listened to the whole hour. My biggest surprise is at how much Wolfram emphasizes the natural language interface, which I’d thought from my own preview was less of a focus–and even the system’s weakest link. In fact, when Rucker asks if there will be a manual offering advanced users a list of expressions akin to those in Mathematica, Wolfram says no, that users are lazy, ontologies never line up, and that the system will figure out what the user means. Of course, I now question my assumption that Wolfram Alpha’s real goal or potential is to act as a service to use in other applications.
At the same, Wolfram says that he envisions “a new field of knowledge-based computing. Imagine a spreadsheet that can pull in knowledge about the entries.” I heard that and then expected him to explain a vision around an API, but he continued to explain that the interface from a spreadsheet would use natural language. I don’t know whether Wolfram has actually thought this through, or if he can appreciate the perspective of an application developer depending on consistency from a software service. Maybe I underestimate him. Needless to say, I’m a lot closer to my initial skepticism again.
What is interesting from a non-technical perspective is that Wolfram sees Wolfram Alpha as doing for NKS what word processors and databases did for Turing’s theory of computation, i.e., proving the value of his grand opus through its consumerization. I’m give him credit for putting his credibility on the line this way, but I think he’s taking a big risk here.
The podcast is a bit rambling, but it’s primary source material–and I found it enlightening enough to devote an hour to it (while writing this blog post). I suggest skimming Rucker’s write-up first, and then deciding whether you have the patience to listen to the whole podcast.