bflbfl

very cool!

@phorese - thanks for the suggestions - great color info

For those who might be interested, I have updated things so that * colors are a little brighter (not right yet, but maybe better for now) * the list is always prepended to, not appended this immediately helps convey the "velocity"; if you mouseover an element in the list, it should pause the insertions (although the counters keep going); when you mouse out, all the "backed up" events slide down. Might be some quirks on browsers with that, but not using anything fancy... this has not been tweaked to work well on mobile browsers yet either.

* there is an option to have it append to the list instead of prepend. Not sure this is needed at all, now that I think about it

There might be 503's sometimes. I don't think it's a quota issue, but just that googledrive can do that under load. I think http code 403 would mean a quota issue. Could be wrong, though. I removed an unnecessary 1MB(!) download that might help.

@edem

It's also at these two places

https://googledrive.com/host/0B_n1OLMaOursX2hsRUhZNElwZmc/

https://googledrive.com/host/0B4G_4-zdiD1gU2NoX285dmxpTXM/

FYI - I saw some weird googlesite errors that might be related to the (moderate) load it's getting at the moment.

The visualization is also on this googledrive site

https://googledrive.com/host/0B_n1OLMaOursX2hsRUhZNElwZmc/

@femto - that's a good point. That occurred to me, but for the moment I decided to stick with the red/green that was already used for the previous visualization. Well sort of, as I actually use d3.rgb.darker on the actual colors because I wasn't satisfied with the original ones, as they looked too light. Even for red/green, I think there's a lot of room for improvement here to make them stand out more.

I'm not sure off the top of my head what would be best for the rg replacement. Maybe some subtle shape differences, etc.

@ahassan - good catch. The geometry data file had them together I guess and I need to go through and split that up.

@arscan - that is really cool. I'd say go for making your own. There were already world meters out there before I decided to make this one. It's a great way to learn all sorts of stuff, imo.

@tectonic - I know what you mean about just a few minutes is all you need to kind of get a feel for things. thx

@a3n - fyi - you can hover on a specific country to see what's going on there

A few initial notes on this are here:

http://www.nowherenearithaca.com/2013/10/visualization-of-wo...

Thanks for the google ref - seems to definitely say it's valid in some cases (e.g., when you expect them to generally stay on the page they come to).

awesome

Didn't remember who he was until I saw the "Inventing on Principle" cover slide - that talk and what he was doing in it blew my mind. It just felt like some kind of breakthrough about how to think about and do things with tools that now make these crazy and useful things possible... with (it seems) lots of cool applications for more efficient and satisfying education/learning.

For those who may come across this, I have added some extra stuff to let you watch how the power method converges (or doesn't converge) to the PageRank vector. Just click the "Explore Convergence" button over there where the PageRank vector is written out.

Very very cool! Those Picasso minimalist sketches have intrigued me as well.

Hi Mike

thanks for taking a look at that. Linear algebra/eigenvector/eigenvalue stuff is useful and usually fairly (re-)approachable.

I am a student of interesting things :)

thx - think I fixed that the other day.

Thanks Mike. I've been playing with small networks as well (partly to confirm things working correctly) - wrote up a few notes on one little one with slightly counter-intuitive behavior - http://www.nowherenearithaca.com/2013/05/google-pagerank-eva...

Hey wfunction

I believe we know exactly what the matrix is, based on the hyperlink structure itself. It's updated and displayed on the vis here based on that, too.

btw, David Austin's article on PageRank is highly recommended : http://www.ams.org/samplings/feature-column/fcarc-pagerank

Hey Mike - can you give me some details on the examples you're playing with? If I get around to it, I was going to see about including some canned examples to maybe show/disprove how things like that work (at least with this version particular of the algorithm). The inclusion of the dangling nodes matrix and that extra all-ones one can make it trickier to predict what will happen.