Comments on: Friction in Human-Computer Symbiosis: Kasparov on Chess http://blog.palantirtech.com/2010/03/08/friction-in-human-computer-symbiosis-kasparov-on-chess/ Articles from the Engineering Group at Palantir Technologies Tue, 24 Jan 2012 09:51:18 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Albert http://blog.palantirtech.com/2010/03/08/friction-in-human-computer-symbiosis-kasparov-on-chess/comment-page-1/#comment-837 Albert Wed, 19 Jan 2011 08:57:50 +0000 http://blog.palantirtech.com/?p=1302#comment-837 Hi Ari, I love the thoughts you gave and the example of the chess challenge. I’d suggest creating two components for the human-computer interaction, also sort of building on Asher’s post. I once wrote a chess game based on the thoughts of Chomsky. To do that, I distinguished the modes of thinking of a chess player (modeling a humble myself), and derived a method for dynamic analysis of the ‘text’ of a game in progress. The solution, made in LISP, could find a ‘mate in three’ solution easily, without calculus like Deep Blue. I used a board as input/output device. From that I derived a sense of analytics that differs widely from common perception. So a suggestion is then to differentiate from both a and h, into an analytic capability that follows the human nature with its intuition, rules of thumb (for instance called A-i) against pure calculus or data crunching. Maybe the h part can be extended to to show the capability to converse in ‘human language’ with the system (H-l for instance). Now how does this lead us to Palantir? The component H-l for me closely resembles the work done with Helpers. These are sort of formalized methods to grasp the complexity of a domain of thought about day-to-day problems and solutions. The component A-i reflects the possibility to have a shorthand, for visualizing the question and the corresponding (re)presentation space, and being able to transmit this to the computerized part of the analytical processing. So for instance, while a number cruncher can provide ‘relevance’ in many background texts (e.g. police filings), a trained analyst will run through a set of ‘diagrams’ and grasp pertinent relations quickly to pinpoint the needle in the haystack along with a corresponding real-life hypothesis. Human agents can ‘fill in the dots’ and see the ‘obvious’ in a way that evades scrutiny; good tools support them to use their intuition. That is just the learning point of the 2005 Freestyle Tournament. By splitting up the function, it is possible to model the amount ‘earned’ with additional appropriate tools. Earned in terms of resolving power of cases, in some instances, the capability to bring to court with evidence. Take for example the large investment made by Palantir in the Revisioning Database or the Ontology. How well is additional money spent to cater for ‘business situations’? In marketing science, functions have been made and tested that calculate the ‘last marginal dollar spent’ on advertising that is effective. Marketing managers use such functions, tailored to the channels, to claim and manage their budgets. By expanding the calculus that Asher developed, a tool evolves that shows up to which amount it ‘pays off’ to invest more in Helpers or other areas and why it is wise to have e.g. an Ontology Board. Hi Ari,
I love the thoughts you gave and the example of the chess challenge. I’d suggest creating two components for the human-computer interaction, also sort of building on Asher’s post.

I once wrote a chess game based on the thoughts of Chomsky. To do that, I distinguished the modes of thinking of a chess player (modeling a humble myself), and derived a method for dynamic analysis of the ‘text’ of a game in progress. The solution, made in LISP, could find a ‘mate in three’ solution easily, without calculus like Deep Blue. I used a board as input/output device. From that I derived a sense of analytics that differs widely from common perception.

So a suggestion is then to differentiate from both a and h, into an analytic capability that follows the human nature with its intuition, rules of thumb (for instance called A-i) against pure calculus or data crunching. Maybe the h part can be extended to to show the capability to converse in ‘human language’ with the system (H-l for instance).

Now how does this lead us to Palantir? The component H-l for me closely resembles the work done with Helpers. These are sort of formalized methods to grasp the complexity of a domain of thought about day-to-day problems and solutions. The component A-i reflects the possibility to have a shorthand, for visualizing the question and the corresponding (re)presentation space, and being able to transmit this to the computerized part of the analytical processing.

So for instance, while a number cruncher can provide ‘relevance’ in many background texts (e.g. police filings), a trained analyst will run through a set of ‘diagrams’ and grasp pertinent relations quickly to pinpoint the needle in the haystack along with a corresponding real-life hypothesis. Human agents can ‘fill in the dots’ and see the ‘obvious’ in a way that evades scrutiny; good tools support them to use their intuition. That is just the learning point of the 2005 Freestyle Tournament.

By splitting up the function, it is possible to model the amount ‘earned’ with additional appropriate tools. Earned in terms of resolving power of cases, in some instances, the capability to bring to court with evidence. Take for example the large investment made by Palantir in the Revisioning Database or the Ontology. How well is additional money spent to cater for ‘business situations’?

In marketing science, functions have been made and tested that calculate the ‘last marginal dollar spent’ on advertising that is effective. Marketing managers use such functions, tailored to the channels, to claim and manage their budgets.

By expanding the calculus that Asher developed, a tool evolves that shows up to which amount it ‘pays off’ to invest more in Helpers or other areas and why it is wise to have e.g. an Ontology Board.

]]>