Filed under: Uncategorized
The wraps are coming off of Wolfram Alpha, and even in infancy it is poetry to my left angular gyrus. A recent review compares it with Google. Many rightly note that the two aren’t really tackling the same problem. The interesting points to note in the review are those hits where the questioner is flooded with detailed data that, among other things, advises him about what dimensions of reliable detail exist.
On one hand, Google’s original mission, to provide information effectively to the world, does encompass this… but then it encompasses many things they do not yet pursue. They have avoided projects that trying to directly answer difficult questions, or to provide more than an algorithmic gloss of public approval on top of public link- and word-association. Google Answers, which I would have expected to stay around forever even at a loss for its direct contribution to the mission, was shelved years ago. Brief forays into public data portals for gov docs in the US and educational materials remain tentative.
So deep access to public data has been better provided by individuals such as Carl M. and by university research projects than by Google. Of their recent initiatives, only their book and catalog scanning projects have really been successful at making public data more available.
On the other hand, the goal of search engines has always been basic search. The field, and its developers and system-gamers, have been working with a limited, transient concept of what it means to capture the desire for knowledge and provide related resources.
For instance, while there are millions of excellent marks or stamps of data quality and fidelity embodied in the organizations that gathered them, how they are labelled, and how one might verify thatthe digital source you are using is in fact the canonical one for the creating org, none of this has been part of ‘search’. The fields of data quality and integrity, org and host authentication, stamping and validation by orgs and evaluating groups and reviewers, and the like have hardly been budged forward over the past decade, despite the tremendous success of ‘search engines’ as a type of service.
So there is no doubt that other essential services are missing, ones which considers ‘providing nuanced answers to questions’ their primary goals, and rely on well-ordered data, on honing in on better answers through collaboration. Data-focused services should have two-way channels with providers and hosts to help them make their data more accessible, and amplifies quality over popularity for relevance when it comes to resolving or informing nuanced questions.
Human-facing elements should make the process of question-asking, as something each of us learns to do hundreds of times a day, trainable / reviewable / repeatable. Questioners, data providers, and intermediaries should all learn how to improve, not just the latter. It should be possible to promote an important query that isn’t being answered, to ask for more direct attention, thereby improving both service and experience. And there should be ways to cluster questions so that similar questions return the same cluster of results, and can accept comments and direct suggestions from participants. Often when asking a question you really want to see a refined and more precise variant that someone else has asked… and they may want to see if the answer has changed since they asked it.
But back to WA. It’s thrilling to me to see a new kind of frame for the ancient oracular challenge, since I had lost hope of significantly different innovation emerging from current notions of ‘search’, or simple human-heavy ‘answers’ services. This promises to address just one facet, but it’s truly a new one.
It’s due out this month, and word is there are O(100) people working on it, so keep your eyes peeled.