holy crap, more math than what i would care to do

That's why I have computers do it for me!

It might be more math than one would care to do, but is it More Than I Needed To Know?

heh eh. popufur.

which i promptly typoed as "popupfur", which sounds as accurate.

Oh my... I'm so confused

yay, as a video game programmer i only clicked on this because it said algorithem!

good job :)

must be really cool to program video games =)

actually its not that great. i hate everyone and everything! lol =]

Meh, you don't need to be a programmer to hate everyone and everything. Just a human being. :)

I approve of this submission :p I always wondered how you assigned popularity weightings especially with regards to sink nodes, makes sense now.

Imma take a look at your sauce!

Lost me at when calculating what D will get :P

Thanks for making it tho ;3 Sounds complex!

Mmmm... Computer science rules all!

Haha, that is quite awesome, always wondered how that worked. Interesting using the 15% of unmitigated outgo rank to non-watchbacks as a balancer, I would never have guessed! :P I wonder how the results would change if the formula could incorporate +favs shown on a given user's userpage modified by the incoming watchers as weight, would be impossible to implement though as way too much data changes for it to be reliable by watchers alone (+watch doesn't necessarily mean +view at any given moment) and would take waaay too much bandwidth off-server, rofl. Not to mention massive balancing flaws that would show up... Just thinking aloud, err, in text :P

Also, I recognize that font, sup iidx system menus. :3 (yes, i know this is a common font, still)

The font? It's just MS Gothic. I don't think it looks too much like IIDX menus. Which version? I always thought it was something like Arial or Helvetica.

It would be fun to "weight" different edges of the graph according to fav count. It wouldn't be TOO hard to "implement" inefficiently, but getting that data would be a pain in the ass. 146K known accounts, 12.7M known watches- potentially 21 billion <userA, userB> fav-count pairs, and that's JUST the accounts I know of! I doubt it'd change much in the end, haha.

Haha, true, so much data makes every change a big deal, and overall its probably a small percent that actually are referred through the fav system and being able to generate an accurate stat :P Would need views, referred through the +fav interface, that then linked back to the userpage to get an accurate read of the generated traffic, that would even just be used for the weight (which would by then be so minuscule it wouldn't matter much rofl). Straight up fav count would be a massive operation in itself like you said (21 billion is insane, O.o ) and it's not like the current system needs changing anyways, heh.

Also on the font, I would always see it in either DD or Gold's boot menu [AC, sup sows] where I would spend hours getting sync perfect by testing a song, re-timebasing, then trying again, though eventually autosync worked well for my USB adapter on an older Dao asc come Troopers :P I used to play alot back then, haven't really done much iidx since win7 beta, though occasionally I drop by sunnyvale golfland (cali, the sirius machine last I checked, dunno if they upgraded) so MEH. Long rambling post is long and rambling, lol.

Oh, that! Yea, that might be the font. It's probably something else like Mincho, who knows.

Haven't been to SVGL in a long time. I'm in the Phoenix area, so Mesa GL is all I have. That place is meh. Used to be cool, but I find that playing at home is just as much fun, now that AC DJT and SN2 are old hat.

Fun fact: MGL's IIDX machine boots up in English Windows XP Embedded (with the standard font) because it's a bootleg. At least it did when it was Gold.

Given Bemani's irregularity with...well, everything, I wouldn't be surprised if its some slightly changed obscure font :P

I totally forgot about MGL altogether, I'm over in the SF bay area so that is actually a bit closer than SVGL I believe, will have to head over there sometime! :P It is true that its pretty much the same as home ac unless the arcade has a ridiculously new version, and even then troopers > emp imo, especially when there's no eA to emulate most of the features in the newer versions anyways. I do like sirius' UI and will have to check out 18 if somebody updated though. ;3 Made me really sad when that ass on 2ch reversed the sirius crack for coin mode, was reaaaally hoping to get co.ws support which was almost complete, ever since that I really haven't played as I <3 stat tracking, lol.

Rofl at the XP Embedded btw. I think the English GOLD that was demoed...somewhere had the same setup, kinda shows how unfinished the build was :P Too bad they never brought iidx/eamuse over, though it might be too niche of a market here and I dont blame bemani, given official export/localization costs. =(

MGL = Mesa, AZ - not Milpitas. I always gotta remember to mention that...

English Gold in Naperville IL was amazing. I have a ton of footage from that on YouTube. It was Japanese XP, I believe. All they did was change the information boxes to be in English. Eeeeeverything else was left alone.

Kinda disappointed it never made it here, but probably for the better. Nobody would have ever played it!

If it's not a secret I wonder how many iterations it takes to reach the convergence for such a complex network?

Suprisingly, not that many. Maybe a dozen?

I expected that... usually convergence is very fast or very slow. It depends on a process... For example for iterated diffussion equation implemented in naive way converge pretty slow... Thank you for reply ^_^

PS: One thing... comment your code when you show it to the public, please... understanding what each line does is simple but understading an underlying idea of a block of code is sometimes hard... it takes unnecessarily much time :)

oh sorry, you put your code in the journal not here :D so it was about the code in your last journal.

Thanks for explaining that o.o It was difficult to follow in the end but I think I understood.