{"items": [{"author": "Patricia", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=332188850150605", "anchor": "fb-332188850150605", "service": "fb", "text": "This is very cool, but I still have trouble understanding it at a glance. What if the arrow also changed color -- green (recent) to red, maybe?", "timestamp": "1330298789"}, {"author": "Mac", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=332253716810785", "anchor": "fb-332253716810785", "service": "fb", "text": "So why does a bus deviate from its expected \"world-time coordinate\"? It might have broken down at its last stop and not moved since. Or there might not be much traffic and thus be traveling twice as fast as normal. Or it might just be traveling at its nominal speed and hasn't updated. A presentation that would appeal to me would be an oblong blob. The long axis of the blob would lie along the route, the center would be denser, and at the location of the bus' nominal position. Maybe a marker dot to make the center easy to spot. The blob would have a gradient fade to 0 opacity at the extremes of the bus' possible positions. Your long arrows adequately indicate the likely range of positions. But maybe a gradient blob would be easier to intuitively grasp at a glance. Make the blob transparent enough at its center to still read all the map data through it. The length of the blob, proportional to uncertainty, could be an indication of the freshness of the data. Or you could use colors. Maybe a pleasing blue fading to a dull gray. Tufte's \"Envisioning Information\" is a towering work on the subject.", "timestamp": "1330307771"}, {"author": "Danner", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=332514853451338", "anchor": "fb-332514853451338", "service": "fb", "text": "Jeff, I have this book, if you want to borrow it. it is pretty, and very useful. I like the line, although a prediction cloud is more intuitive.", "timestamp": "1330356579"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=332524626783694", "anchor": "fb-332524626783694", "service": "fb", "text": "@Walker, @Danner: I've read the book; I agree it's quite good.
I have two problems with a prediction cloud. The first is that it gets farther from the underlying data. The line connects the best-guess position to the last-known one; I know how to think about that. The second is that I don't currently have the data to say what the probability distribution is for the bus's position. All I know is it's last reported position and the next few predicted stop arrival times. Probability would mean collecting a lot of data from nextbus, or them opening up their api to export it.", "timestamp": "1330357803"}, {"author": "Rick", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=335342759835214", "anchor": "fb-335342759835214", "service": "fb", "text": "Jeff, how about a checkbox that gives users the option of showing probability tails (like your blue arrows) or turning off this functionality. I think many users would find it distracting/confusing, while others would appreciate it. I think many users would just want to see your best estimate as to where the bus currently is.", "timestamp": "1330771673"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=335387439830746", "anchor": "fb-335387439830746", "service": "fb", "text": "@Richard: I used to have lots of checkboxes on the site for optional features and I took them off because they made it confusing to use and I don't think anyone was using them. I'd like to set the site up to be good by default for most people, and if most would not like the blue tails I should have them off.", "timestamp": "1330779555"}, {"author": "Joshua", "source_link": "https://www.facebook.com/jefftk/posts/332129836823173?comment_id=335453553157468", "anchor": "fb-335453553157468", "service": "fb", "text": "How about a new indicator when the location is very recent. People are used to new/unread indicators. It wouldn't tell you how old a location a prediction is based on, but it would alert you that some predictions are current and others aren't. It would also let people know how frequently positions are updated.", "timestamp": "1330788431"}]}