Ward via matrix 
|
15 Aug 2025 04:32 PM
@Ralf: I remember the extensions you made to Speed Bot, a Markov Monkey inspired attempt to enlarge the Monkey's scope beyond a single site. The "speed" came from approximating link resolution without fetching pages to examine the priorities present in the journal. This is a fine point that the Java implementation got right.
Is your fedwiki-smalltalk running inside GT with all of its analytic resources? You could make a collection of "SuperMonkey" models that more realistically model various reading strategies and then set them loose on real sites. You also made SpeedBot interactive, dare I say "conversational". When it paused, you could direct it to look further from a location (more correctly from a point in its recorded state) and then set it off again. Let's call this man-machine cooperation "Monkey Chat". Then with all the representational power of DMX you will be set to pass informed judgement on hypertext.
See 'Traversing graphs with DeepTraverser'. docs