Hacker News new | past | comments | ask | show | jobs | submit login
Inspired by XKCD:903, Wikipedia steps to philosophy (ryanelmquist.com)
237 points by HistoryInAction on May 26, 2011 | hide | past | web | favorite | 131 comments

This is no joke: 21 clicks from Kevin Bacon to Philosophy. Welcome to the Internet equivalent of a child repeatedly asking, "why?"

This shows a that the program linked is broken. The steps taken, are:

1. Animal House 2. ....

Going to the actual Kevin Bacon Wikipedia page, Animal House is clearly italic. The First non-italic, top-level (wrt parens) is Golden Globe

I expect the program is working off an out-of-date database. Hitting wikipedia directly would be rather bad form.

It took me 30 steps to do it manually, following the correct rules.

Haha ya, it's crazy. I went to Wikipeida.com, clicked on a random link (Ratko_Mladić), and from there it took me 26 clicks of first links to reach the Philosophy page.

When I first read the comic the other day, I went to the page in the comic (Spark Plug) and tried it -- didn't work, I ended up in a cycle between 3 math-related pages. Oops.

likely a result of subtle vandalism by other xkcd readers.

minor correction: 20 clicks

Three, actually. Kevin Bacon > Meme > Philosophy.

This statement shows a gross misunderstanding of the rules.

Meme is quite far down the page. It is well after many links that are neither italicized nor inside parens. The statement whose veracity is being checked says First link (plus some other parameters).

A few years ago at Railscamp in Australia, Andrew Grimm proved this exact outcome. His code is here: https://github.com/agrimm/philosophy-dump-parser and https://github.com/agrimm/philosophy-navigation.

Hah. The hypothesis is false, but the script already has that covered via loop detection... nice :)


uh oh... found a loop

panhard -> auverland -> panhard.

So far the longest trail I've found, at 25 steps, was from 'ED209' via pottery, through minerals, states of matter, knowledge, finite sets, mathematics... that's one heck of a wikitrail.

[edit] Scratch that, 'Horst link' is longer via one step, going via transport, commerce, San Juan de Dios Market, Mexico, Romance languages, Precambrian, Chronology, etc...

30 for "Howard Hughes"

uh oh... found a loop

mathematics -> quantity -> property_(philosophy) -> modern_philosophy -> western_europe -> europe -> continent -> landmass -> landform -> earth_sciences -> science -> knowledge -> fact -> information -> finite_set -> mathematics.

Makes sense for mathematics to be a loop since it's a self contained system.

Hmm. I just tried "Howard Hughes" and got only 15 steps.

Business_magnate -> Philanthropy -> Ancient_Greece -> History_of_Greece -> Greeks -> Nation -> Sovereign_state -> -> State_(polity) -> Social_sciences -> Field_of_study -> Academia -> Community -> Interacting -> Action_(philosophy) -> Philosophy

Has the script been tweaked already?

modern_philosophy -> western_europe

Looks like "philosophy" is the first link on the "Modern philosophy" page (and was last week when I first tried this).


I wonder how many Wikipedia edits this week are going to be to undo loops.

Me, I'd be more inclined to create them >_>

"Sushi" leads to a loop, too:

sushi japanese_cuisine japan japanese_language arabic_numerals symbol numeral_system mathematical_notation uh oh... found a loop

Its supposed to stop at philosophy..

Heh. "Philosophy":

uh oh... found a loop

philosophy -> existence -> sense -> organism -> biology -> natural_science -> science -> knowledge -> fact -> information -> sequence -> mathematics -> quantity -> property_(philosophy) -> modern_philosophy -> philosophy.

F-111 is 45 steps.

Now 32.

Playing games with an ever evolving game board is... interesting.

Now 36

I found a loop for Jersey Shore. I guess that sounds about right though.

"Beans" is 28 steps..

found another loop, on my first try nonetheless:

Genghis_Khan -> Mongolian_language -> Mongolia -> Mongolian_language

There's no loop there:

Genghis Khan -> Borjigin -> Clan -> People -> Human -> Taxonomy -> Science -> Knowledge -> Fact -> Information -> Sequence -> Mathematics -> Quantity -> Property (philosophy) -> Modern Philosophy -> Philosophy

Mongolian language is in parentheses, so you don't end up in the infinite loop.

So far, the only loop I can see is Panhard, though I'm sure someone will edit the page so that no longer happens. I think the stub pages have a far better chance of defying this law than any other page, though, so it's probably best to start there if you want to find more loops.

Font loops, at least according to the rules (ignoring italics and parens).

If it finds a loop, can't it go and explore other paths?

Please spend 38.9 seconds trying to understand the very very very simple algorithm that says: Follow the first link that... It doesn't say there are alternate choices. If you went with a different link than the first, it would no longer be the first.

Well, it is Wikipedia. The program _could_ go to Wikipedia, edit a page in the loop, save, and continue :-)


       1. Recursive
       2. Recursion

    uh oh... found a loop

    recursion -> recursive -> recursion.

Recursion, self-similar, mathematics, quantity, property, modern philosophy, philosophy

Assuming you don't include the links in the "lacks inline citations" box on the recursion page.

Theres a Wikipedia page on it, it works on 93% of all articles.


Yes. It dates back to 2008. Interesting that someone got bugged by it that time and just now it took this dimension.


19 steps to philosophy

I wonder if we could say everything leads to science. I also reached philosophy via science.

Ratko Mladic (randomly chosen from wikipedia homepage)

        Army of Republika Srpska
	Use of force
	Conflict resolution
	Italic languages
	Indo-European languages
	Language family

emacs doesn't traverse through science but does hit computer science


Main_Page is currently taking 19 steps too. (According to the program—other comments here point out that it may be ignoring the comic's rules.)

I tried it with Iraq, but the page parsing is slightly wrong. Got this result.

Iraq Arabic_language Languages Human Precambrian Eon_(geology) Chronology Time Measurement Magnitude_(mathematics) Property_(philosophy) Modern_philosophy Philosophy 12 steps to philosophy

But the link to Arabic is in parentheses. The first non-parenthesesed link is to Western Asia.

An article[1] from the Xamuel.com's blog predates the xkcd stripe by a few days and states the sames things but with mathematics instead of philosophy. Also their's some fun with second links.

[1] http://news.ycombinator.com/item?id=2573038

And the version using philosophy appeared on Reddit before the comic as well: http://www.reddit.com/r/philosophy/comments/hgwbe/on_wikiped...

Two problems. It favors links in the sidebar over article links. I was in "Human" and your script picked out Pre-Cambrian when it should have been Taxonomy.

Second, capitalization matters. It couldn't find "the black keys" but it found "The Black Keys".

It's pretty obvious in hindsight when you think about it. You start with something specific and then get more and more vague until you hit Philosophy.

It totally worked for me, I tried it with Sausage and Cradle of Filth. I thought it was a joke, it blew me away.

But then I found that "Osama Bin Laden" crossed "philosophers" and then ended on a loop between "reason" and "rationality".

EDIT: I just tried "Osama Bin Laden", the steps it takes seems to be using a slightly different Wikipedia than I see.

I tried OOP, eventually arriving at Philosophy -- OMG, it works :))

That's 'cause Philosophy covers it all:

"Philosophy is the study of general and fundamental problems, such as those connected with existence, knowledge, values, reason, mind, and language." --http://en.wikipedia.org/wiki/Philosophy

Too bad there's no Philosophy in http://xkcd.com/435/

Especially since a common path is to reach philosophy by first visiting the mathematics page:


Wow, someone just modified Modern Philosophy (now first link is Western Europe), Science and/or Knowledge pages, which makes most previously things tried by me kick into a 25 loop. I'm wondering if it's intentional.

Most things were getting stuck in a loop around "Indo-European languages" last time I checked. Someone is definitely playing silly buggers.

I was tempted to see whether I could somehow engineer an all-roads-lead-to-goatse situation, but I couldn't think of a plausible pathway.

Just hijack Philosophy.

Too easy. The idea was to create an entirely plausible pathway of links that nobody could complain about individually (so they wouldn't get reverted) which nonetheless eventually led to http:en.wikipedia.org/wiki/goatse

I mean, create that plausible pathway with philosophy as its root. Since so many articles already lead to philosophy, it's a good starting point.

You don't reach Philosophy from Philosophy.

Hey, I did!

Did it using the software and then checked wikipedia by hand:

Philosophy -> Existence -> Sense -> Organism -> Biology -> Natural_science -> Science -> Knowledge -> Fact -> Information -> Finite_set -> Mathematics -> Quantity -> Property_(philosophy) -> Modern_philosophy -> Philosophy

15 steps to philosophy

The "Quantity" article did not contain the link to Property_(philosophy) before and it was added purely for the sake of this game. See the talk page for more details.

Personally, I think its sad that people are editing wikipedia purely for the sake of making games work out.

Yes, you do. 15 steps.

Related: http://TheWikiGame.com (I built this app, it's made with using django, xmpp, and almost every datatype that Redis provides)

Modern Philosophy now goes to Western Europe rather than Philosophy. And since everything seems to go through that, nothing will reach philosophy anymore.

I did this a while ago with node.js: http://blago.dachev.com/wikidrill


Interesting that the first three i tried took about 17 hops and got funnelled through the "Life" entry.

corndog sensimilia halitosis

This took a few less hops and stayed in 'techne': voip.

Would love to see these searches graphed.

Years ago - one of my favorite sites was Everything2 (still up) - it was run on the slashdot engine. The fun of it was to follow the associated links at the bottom of the entry to see where it would take you.

Slashdot used to link to Everything (back before it was Everything2) as a sort of instant-dictionary for tech terms. Stories would have something like "RSS 2.0(?) and Atom(?) proponents are squaring off..." with the question marks going to the relevant Everything node. Most of the time, the terms wouldn't be defined until the story went live, and then they would be in short order.

It was a neat symbiosis, especially back before Wikipedia existed. However, they were separate sites. Both Slashcode (which runs Slashdot) and the Everything Engine (which runs Everything2) are written in Perl. Everything was never run on Slashcode, though, AFAIK.

E2 was wonderful for links teleporting you off into completely unrelated and equally fascinating subjects - I lost many an evening in that manner :)

I just spent about 20 minutes trying to get anything more than 20.

Wikipedia has the answer, as to not spoil here's a link. http://en.wikipedia.org/wiki/Wikipedia:Get_to_Philosophy#Cha... I like the last loop on the article not crossed out.

Wikipedia:Get_to_Philosophy is not what xkcd is describing, though. That page seems to allow any link to be chosen. xkcd says, "click on the first link in the article text not in parentheses or italics".

For example, Wikipedia:_Get_to_Philosophy provides the example, Optimum "L" filter -> Butterworth filter. But the first link in 'Optimum "L" filter' is "Athanasios Papoulis".

Same here. I finally got it with the word "Coprophagia". 21 steps. I feel oddly proud.

(It came from a conversation some friends and I had about kopi luwak, just to explain why I tried this obscure word).

Nice tool :-) It seems to also take words between parentheses though I think. For example on http://en.wikipedia.org/wiki/Iceland I would expect it to follow European not Icelandic.

Ironically, it doesn't work from "Spark Plug".

Strange, it did for me yesterday.

Edit: Modern Philosophy seems to be missing the link to Philosophy, which should be there according to the edit history (http://en.wikipedia.org/w/index.php?title=Modern_philosophy&...).

Apparently some kid was having fun vandalizing the page to break this. The page is semi-protected now, so it should be ok from now on.

Funny, but the attractor for Conservapedia seems to be "nation".


I suspect that's also one of the most common non-stop words on "The Colbert Report".

By extension, you also will always end up at ... Existence Sense Organism Biology Natural_science Science Knowledge Fact Information Finite_set Mathematics Quantity Property_(philosophy) Modern_philosophy

Bastard beat me to it!

Friend of mine was graphing out all of the connections (using Graphviz) and found some interesting patterns. Ultimately you can say the same thing for "Science", "Knowledge", and even "Mathematics".

Found a loop: Start at Reiki

reiki spiritual_practice spirituality reality being eastern_philosophy islamic_philosophy islamic_studies islamic_history history george_santayana madrid spain balearic_islands catalan_language uh oh... found a loop

spain -> balearic_islands -> catalan_language -> spain .

So this game is traversing a cyclic graph. It should be possible to get around this by keeping track of visited links, and not repeating our clicks.

Now is Wikipedia a connected graph?

'music', 'art', 'war' are examples of non-esoteric terms that don't end up at philosophy.

'libertarianism' doesn't end up at philosophy either :)

Though I'm noticing that even before 'mathematics', all my meme-true queries are going through 'science'

also... 'philosopher' doesn't go through via the script page, but on the actual wiki it's redirected straight to 'philosphy'.

Fun page, nice work.

Are you sure you're doing it right?

music -> art -> sense -> perception -> awareness -> consciousness -> mind -> panpsychism -> philosophy

war -> conflict -> social behaviour -> physics -> natural science -> science -> knowledge -> facts -> information -> sequence -> mathematics -> quantity -> property -> modern philosophy -> philosophy

I'm just going off what the linked page gives me.


    catalan_language (loop to spain)

    military_history (loop to war)
it seems that the script in the main article isn't using the same pages we are

The tool is broken. It's picking up links from the side bar instead of from the main article, since those links show up in the source first.

It's not funny if you edit the articles and remove links to force it to be true.


It seems to take a while to search; are you querying wikipedia each step instead of a DB snapshot?

Last time I downloaded Wikipedia, it was 4.5GB. If I were to knock up this hack, I would definitely scrape pages instead.

On-demand page scrape + memoisation is almost certainly a win here. Even if thousands of people are hitting this, a lot will choose some of the same queries (I'm sure Kevin Bacon and xkcd and philosophy are in there a bunch), especially in the tails of the paths (Latin, Mathematics, ...)

That "Wikipedia Trivia" is really intriguing! Manually tested and verified it.

I know I'm being pedantic here, but I'm pretty sure you haven't verified it. To do that, you would have to visit every Wikipedia article.

I'm quite surprised at how many steps it takes to get from Isaac_Newton to Philosophy (31). I'd expected that would be a quick one. I guess that's just a quirk of the "first link" rule though.

Justin Bieber: 18 steps. I don't know if it is a good thing or not.

The current version appears to dislike unicode characters.

http://en.wikipedia.org/wiki/Petar_Dobrovi%C4%87 causes an error.

Excellent. Saved me some clicking resulting in more efficient use of procrastination time!

I hit loops a few times - you could drop to the second link in those cases instead of stopping completely.

If you drop out of loops, you will reach every article eventually. (Unless you are ending in a cul de sac.)

If the draw to "philosophy" is as string at it seems you should end up there without scanning too many articles. You'd need some sort of fixed cap to avoid the few cases where you'd end up scanning thousands of articles, of course.

Appears you almost always get there via Mathematics though.

I find fascinating that if you end up on "Mathematics_education" than you have no chance of getting to "philosophy". (or "physics" is enough, apparently)

Can anyone find a path that doesn't end with:


Barrack Obama:

barrack_obama List_of_Presidents_of_the_United_States United_States_Constitution Supreme_law State_(polity) Social_sciences Field_of_study Academia Community Interacting Action_(philosophy) Philosophy

cessna_172 Fixed-wing_aircraft Aircraft Vehicles Bicycle Human-powered_transport Transport Cargo Commerce Buying San_Juan_de_Dios_Market Guadalajara,_Jalisco States_of_Mexico Mexico Spanish_language Romance_languages Indo-European_languages Language_family Language Human Precambrian Eon_(geology) Chronology Time Measurement Magnitude_(mathematics) Property_(philosophy) Modern_philosophy Philosophy 28 steps to philosophy

Xkcd succeeded in making me waste 20min yesterday clicking links on Wikipedia checking whether it was actually true. I wish I had had this tool then!

On the other hand, would you have spent less than 20 minutes playing with the tool?

From my limited clicks, it seems like both science and mathematics are even faster to get to in this method than philosophy.

1)ruby_on_rails 2)Open_source 3)Philosophy

I tried and I end up in an infinite loop between science, philosophy, knowledge/epistemology, and language.

"philosophy" is the halting condition

the routes should be affected not by selecting the first link on the page but a weighted probability based on the number of clicks each link on the page gets to the other in-subject links in the body.

also sometimes it counts links in sidebars as first, i think that's unintentional.


Douglas Adams is not 42 links away from Philosophy, sorry :(

Does this mean the world is about to be blown up by the Vogon's?

Sequence, Information and Fact seem to be big attractors

I wonder what the attractors are for Conservapedia?

Why does searching for cannabis just reload the page? I do admit that when it came back and the text entry box was empty I'd already forgotten what I was trying to search for. It wasn't until the next joint when I thought of trying it again that I recalled it had happened previously.

I got stuck in an infinite loop starting from Firefox.

And xkcd itself is "18 steps to philosophy"...

I don't think it works correctly though... it uses "Author" as the first link whereas the first link in the text of the article is "webcomic" ("Author" is in the summary pannel on the side)


The same happens for cities or locations, where there is usually a "Coordinates" in the upper right corner. Still, I did it manually in text bodies, and you eventually end up to Science or Mathematics, which leads to Philosophy.

It seems to be taking the right panel first. Same thing happened to me with Trade[1], leaded to San Juan de Dios Market


So is "Potato" and "Caen", and the majority of other things I tried. For Godwin points, "Hitler" is only 14 steps.

It worked when I tried it with Power Rangers.

Anyone try generating a map of Wikipedia?

Hmm, doesn't work for cryovolcanism.

This image shows how things end up in Philosophy, by XKCD forumite arotenbe: http://forums.xkcd.com/download/file.php?id=29727&mode=v...

Looking at these stats shows how viral this meme/trivia got:




I posted these stats on a FB status a couple of days ago because of the sudden spike. Now that sudden spike looks small compared to today's sudden spike. Funny how there's different levels of viral, almost akin to different levels of infinite. What would happen if this hit mainstream media?

On that chart, Greeks points to to other pages. How can Greeks have two first links?

That was my question, too. There doesn't seem to be a disambiguation that would apply, either. "Greek Language" shouldn't point to "Indo-European languages", it should point to "Greeks". I can't figure out how that went wrong, either, but it is interesting to see a cycle in an acyclic graph. :)

Schroedinger wrote that page.

As of 2008, the center of English Wikipedia is the article 2007. From that article, it takes on average 3.45 clicks to get to any of the 2,111,479 articles reachable from it. Disregarding all of the articles that are just lists, years or days of the year, the "real article" closest to the center is United Kingdom, at an average of 3.45 clicks to anywhere else.

from http://en.wikipedia.org/wiki/Six_degrees_of_separation#Wikip...

>the "real article" closest to the center is United Kingdom

I find this absolutely fascinating. Consider that the reason we're at the centre of the world map is because we had the power to choose to put ourselves there, at the point in history when the maps were being drawn.

Is that imperial influence, long since waned, still echoing across the years, visible in the wikipedia graph?

OR, it's just because they used the English wikipedia for it

Hah, good point. I wasn't trying to be triumphalist, though I admit it certainly reads that way.

I got stuck on loops each time I tried it, and figured he made it up. Damn.

15 stepf from Philosophy to Philosophy ..

what are the implications?

Also try turning off spell check or writing something significant on paper and see how well we do these days.

Actually spell check, has made me a better speller. Of course in the old days when the apple ][ dominated the personal computer space you had to save your word processing file, exit the word processor, switch floppy disks, start the spell checker, switch the disk for the dictionary disk, run against your file, approve/reject suggested changes, save the file, exit the spell checker, switch to the floppy disk with the word processor on it, start the word processor load your file, find the things the spell checker flagged as wrong but couldn't find a correct word for and fix those.

Seriously, it was just easier to get better at spelling.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact