Wolfram|Alpha - the future of web search technology?
March 8, 2009 2:17 PM   Subscribe

Could Wolfram Research's (creator of Mathematica) Wolfram|Alpha be the future of web search technology?
But what about all the actual knowledge that we as humans have accumulated?

A lot of it is now on the web—in billions of pages of text. And with search engines, we can very efficiently search for specific terms and phrases in that text.

But we can’t compute from that. And in effect, we can only answer questions that have been literally asked before. We can look things up, but we can’t figure anything new out.

[...]

It’s going to be a website: www.wolframalpha.com. With one simple input field that gives access to a huge system, with trillions of pieces of curated data and millions of lines of algorithms.
What does it actually mean? TechCrunch breakes it down:
It doesn’t simply return documents that (might) contain the answers, like Google does, and it isn’t just a giant database of knowledge, like the Wikipedia. It doesn’t simply parse natural language and then use that to retrieve documents, like Powerset, for example. Instead, Wolfram Alpha actually computes the answers to a wide range of questions — like questions that have factual answers such as “What country is Timbuktu in?” or “How many protons are in a hydrogen atom?” or “What is the average rainfall in Seattle?”
True Knowledge, launched in 2007, uses similar technology. Powerset, yet another competitor, was aquired by Microsoft in 2008 but has yet to live up to its promise.
posted by Foci for Analysis (81 comments total) 16 users marked this as a favorite
 
It appears to be an expert system.

Meh.


Meeeeeeeeeeeeeeeeeh

How is this any different than what has come before?
posted by zabuni at 2:20 PM on March 8, 2009 [1 favorite]


Wolfram's arrogance is breathtaking, even for old hands in the sciences who are used to arrogance in their colleagues. Mathematica is quite good, New Kind of Science is mostly garbage with some interesting ideas buried in an un-peer-reviewed academic tome lacking any citations. Full credit to Wolfram's company for actually putting product out, looking forward to trying the website. But there's hype, and then there's delivery.
posted by Nelson at 2:22 PM on March 8, 2009 [8 favorites]


"How can the net amount of entropy of the universe be massively decreased? "
posted by Iridic at 2:23 PM on March 8, 2009 [20 favorites]


What happened to Hart?

/obscure nerdity
posted by Caduceus at 2:31 PM on March 8, 2009 [8 favorites]


'How did life begin?'

'Was Jesus the son of God?'

'Do you really get a pack of virgins in the afterlife, if you're a good Muslim?'

'If God had a penis, could he bend over far enough to blow himself, and if so, would his orgasm destroy the space-time continuum?'

'How many Republicans does it take to screw in a light bulb?' (They don't, they screw in mens' bathrooms at airports.)

All these questions, and many more, will be answered with the release of Wolfram Alpha!
posted by jamstigator at 2:35 PM on March 8, 2009 [10 favorites]


I'm holding off judgment until I see how it answers this classic.
posted by Derive the Hamiltonian of... at 2:35 PM on March 8, 2009 [3 favorites]



"How can the net amount of entropy of the universe be massively decreased? "

Well, Wolfram could stop talking about himself.
posted by jamjam at 2:41 PM on March 8, 2009 [5 favorites]


Is there a God?

Answer.
posted by adipocere at 2:45 PM on March 8, 2009 [4 favorites]




^ hardy har har har to the snarkfest above.

This is exactly the sort of advance that I've been expecting if not exactly waiting for, an AI system that integrates search with semantics.

"What was the population of Tokyo and Yokohama in 1970"?

"How many children did Harry S Truman have?"

"What was the last country to declare war on Germany in WW2?"


These are simple questions and to get the answer right is non-trivial but not theoretically impossible.

I had expected Google to get there first but props to wolfram for trying and I cautiously await the chance to try this out. It is the holy grail of 20th-century AI research.
posted by troy at 3:05 PM on March 8, 2009 [3 favorites]


If not as useful as Google, I hope its results are at least as entertaining as Cuil.
posted by cj_ at 3:09 PM on March 8, 2009 [1 favorite]


Google actually gives the answer to each of those questions in the first result.
posted by unknowncommand at 3:10 PM on March 8, 2009 [7 favorites]


The questions that troy just asked, I mean. The third one is kind of crappy, but it definitely points you in the right direction. I was seriously impressed, actually.
posted by unknowncommand at 3:14 PM on March 8, 2009 [1 favorite]


I don't see much new here yet. Call me when you've got some scheme where the search algorithms themselves are maintained by the users themselves via a wiki & genetic programming hybrid.
posted by jeffburdges at 3:19 PM on March 8, 2009 [1 favorite]


If this reinvents web search the way A New Kind of Science upended science, we're in for a wild ride.
posted by jayder at 3:21 PM on March 8, 2009 [8 favorites]


I keep entering "When did SkyNet become self-aware?" and it keeps saying "Launching May 2009."

I'm a little concerned.
posted by 0x029a at 3:38 PM on March 8, 2009 [13 favorites]


Actually, this is good. Not new, no. And yes, it sounds like an expert system. But think about it. An "expert system on everything" built during the Google Age? Yes, universal expert systems have been built before, but this one is being built in a time period when people have a greater chance of caring, and therefore it may get more use, and therefore it may actually go somewhere this time.

Notice that Wolfram says that this won't parse natural language, which means most of the questions mentioned above are probably phrased incorrectly. It probably requires questions to be posed in some vaguely Lisp-like syntax (I mean, if you've seen Mathematica, you can probably guess that much), and I doubt questions about life, the universe, and everything can be phrased using such constructs. That's obviously not what he's going for.

As a better example of what it is I think he's really trying to achieve: I frequently discover when I'm working on some computer algorithm or something that I need to know whether there's an equivalence between one kind of thing and another, or what the equation is to solve a certain problem, and those questions can be really hard to answer just by searching through math-oriented web pages, even if the answer is something actually known to experts. My life would be easier if I could just ask something like:

(formula What? acceleration angular-momentum)

And immediately get back an equation relating the two quantities, for example.

So let's just wait and see, and hope to the holy name of Cthulu it doesn't involve cellular automata somehow.
posted by Xezlec at 3:38 PM on March 8, 2009 [1 favorite]


Kill it before it Singularities!
posted by Artw at 3:40 PM on March 8, 2009 [1 favorite]


I don't think that's a verb. ("Singulates"? "Singularitizes"? "Singularitificates"?)
posted by nebulawindphone at 3:47 PM on March 8, 2009 [2 favorites]


Singulossifies, actually.
posted by Flunkie at 3:49 PM on March 8, 2009 [1 favorite]


Does it speak any languages other than English?
posted by moonbiter at 3:49 PM on March 8, 2009 [2 favorites]


A New Kind of Searching
posted by DU at 3:59 PM on March 8, 2009


Needs more butler analogies.
posted by Artw at 4:03 PM on March 8, 2009 [2 favorites]


Also Roger Penrose and Ray Kurtzweil should totally do search engines of their own.
posted by Artw at 4:04 PM on March 8, 2009 [2 favorites]


I'll be looking forward to this.

One thing about Google: it just works. Generally, you get what you need in the first page or two. The same can not be said for most other search engines - this is why people keep coming back to Google. It just works.

If the same can be said for this new search engine, wow. My concern is it will be an impressive technology, moving things forward, but not read for prime time, it won't "just work". In fact, he even says it's still early beta and only in a certain fields etc.. so, I don't understand all the anti-hype when there has been no hype other than a long term vision of what could be. The May release is certainly not that.
posted by stbalbach at 4:17 PM on March 8, 2009


Needs more garlic.
posted by RoseyD at 4:27 PM on March 8, 2009 [1 favorite]


Someone needs to search Uranus for Klingons. Heh-heh. Heh-heh.
posted by ornate insect at 4:30 PM on March 8, 2009


Does it speak any languages other than English?

Yes, Esperanto.
posted by ryoshu at 4:31 PM on March 8, 2009 [2 favorites]


Doug Lenat ... paging Doug Lenat ...
posted by RobotVoodooPower at 4:31 PM on March 8, 2009 [1 favorite]


Use in theory:

"What was the population of Tokyo and Yokohama in 1970"?
"How many children did Harry S Truman have?"
"What was the last country to declare war on Germany in WW2?"

Use in actuality:

"How is babby formed?"
posted by Hovercraft Eel at 4:36 PM on March 8, 2009 [4 favorites]


Until someone creates a search engine that collapses the space-time continuum and knows what you want before you do, I'll stick to using divination (Tarot, I-Ching, Magic 8-Ball) for all queries more pressing than those like "what is the population of Albany?"
posted by ornate insect at 4:38 PM on March 8, 2009


INSUFFICIENT DATA FOR A MEANINGFUL ANSWER
posted by monju_bosatsu at 4:41 PM on March 8, 2009 [6 favorites]


Natural language processing isn't actually that hard. What's interesting is that when these "super-uber AI systems!" come out, typing their example queries into Google usually gets you a pretty good answer.

Also, Google does this on some queries. For example, if you Google "population of Georgia" the first result isn't a page, it's an answer:
Georgia — Population: 4,630,841
According to http://www.intute.ac.uk/sciences/worldguide/html/890.html - More sources »
Google's method of slowly phasing in answers that to questions it's capable of answering is probably better then trying to do it all at once.

Now maybe this thing could do something like interpolate the data from 1970 and 1980 to give you an estimate of the population in 1975. That would be kind of cool. The pieces to do this kind of thing are out there, but it would be such a huge project and not worth all that much since there aren't really that many people who need such a sophisticated search system. It's not surprising that the person who would actually come with would be an independently rich egomaniac who can spend years working on projects for his own entertainment.

Assuming it works.
posted by delmoi at 4:41 PM on March 8, 2009


Xezlec: Notice that Wolfram says that this won't parse natural language, which means most of the questions mentioned above are probably phrased incorrectly.

You're right. He very humbly says that, instead of parsing natural language, his "true computational knowledge engine" will compute knowledge itself.

I'm guessing the first thing he's planning on asking his magic machine will be: "oh supreme supercomputer of all knowedge: now that we've got you, well, we're just curious, but how do we write computer programs to parse natural language at Turing-capable levels?"
posted by koeselitz at 5:21 PM on March 8, 2009


I keep hoping that Wolfram will be really famous some day so that all my college job stories about his personal archivist, the months of Photoshop work spent on his bio pic to or 3am rants to the web team that turned out to be an unplugged ethernet cable at his house will be of use in general conversation.

Until then, yeah, Wolfram will replace Google like Publicon replaced LaTex and Quark Xpress.

Or like Cuil killed Google a year ago.
posted by Gucky at 5:39 PM on March 8, 2009 [9 favorites]


Let’s say we succeed in creating a system that knows a lot, and can figure a lot out. How can we interact with it?
posted by anarcation at 5:47 PM on March 8, 2009


My daughter, when she was very young, once asked me, "Why are there so many Ewoks in Star Wars?"

When your AI can answer that question, it will be complete.
posted by Xoebe at 5:50 PM on March 8, 2009 [5 favorites]


How can you use my intestines as a gift?
posted by fleetmouse at 5:54 PM on March 8, 2009


Natural language processing isn't actually that hard.

Um. What? Please explain. And if you've managed to develop a working, reliable natural language parser, by all means let us know about it.
posted by Xezlec at 5:55 PM on March 8, 2009 [2 favorites]


Needs more butler analogies.

First release candidate codenamed 'Alfred'?
posted by mannequito at 5:55 PM on March 8, 2009


You're right. He very humbly says that, instead of parsing natural language, his "true computational knowledge engine" will compute knowledge itself.

That's not at all an unreasonable claim. "Computing knowledge" sounds like it just means taking facts and plugging them into other facts. It's pretty easy and it's been done.

And everyone still keeps talking about this like it's an AI or parses English queries. I'll just say it once more: these appear to be misunderstandings. It sounds like an expert system, which is basically a type of fancy database that can retrieve information with complex structure (in other words, the whole point is to answer questions more involved than "what's the capital of New Mexico?"). It's not magic, and it could be useful.
posted by Xezlec at 6:05 PM on March 8, 2009


Just like the genie and the three wishes:

1. "What is the best wish, for my purposes, to make?"
2. That wish.
3. (Saved for undo.)
posted by aeschenkarnos at 6:09 PM on March 8, 2009 [3 favorites]


The only problem with this service will be couching my search terms as cellular automota rules.
posted by DU at 6:19 PM on March 8, 2009 [1 favorite]


Ahhh, who cares, I've already found all the boobies on the internet.
posted by jonmc at 6:23 PM on March 8, 2009


Ask Metafilter's going to need a new business model!

Actually, I wonder if the tech involved could be run against a limited dataset (like a Google appliance kind of deal). Seems like it would be something that companies with huge knowledgebases, or that have an interest in knowledge management inside their firewall, would be keen to get their hands on.

Being able to reliably run natural-language queries against the AskMe database would be pretty cool.
posted by stavrosthewonderchicken at 6:50 PM on March 8, 2009


Ahhh, who cares, I've already found all the boobies on the internet.

But with this new search engine, you could find out how many boobies are on the Internet. Then you could periodically ask again, and when the number changes, you could go ask for the difference.

Wolfram Research: The The Power to Keep Your Boobies Collection Current
posted by secret about box at 7:06 PM on March 8, 2009 [1 favorite]


Could Wolfram Research's Wolfram|Alpha be the future of web search technology?

No.
posted by erniepan at 7:14 PM on March 8, 2009 [1 favorite]


We've had this for nearly the entire history of the Internet. How much wood would a woodchuck chuck...
posted by Ella Fynoe at 7:23 PM on March 8, 2009


Instead, Wolfram Alpha actually computes the answers to a wide range of questions — like questions that have factual answers such as “What country is Timbuktu in?” or “How many protons are in a hydrogen atom?” or “What is the average rainfall in Seattle?”

I actually put that into google within the past month. Even without clicking on any links I got my answer from reading the third result. The reason that happens is, assuming the question was formed properly, someone somewhere has formed the response to the question and google finds the exact match. So what's new with this? We've already discussed the semantic web (still largely unrealized). This is really just the dumbing-down of (re)searching.

Unfortunately, "...demo his new online service — Wolfram Alpha (scheduled to open in May)" and "I will respect that and not give a visual description of exactly what I saw." So, we really have nothing to discuss, do we?

[but I sure like your enthusiasm!]
posted by sluglicker at 8:56 PM on March 8, 2009


Does it speak any languages other than English?
Yes, Esperanto.


And Vonish...
posted by compound eye at 9:01 PM on March 8, 2009


This is really just the dumbing-down of (re)searching.

Sorry. That was an arrogant thing to write.
posted by sluglicker at 9:01 PM on March 8, 2009 [2 favorites]


It could be an expert system, but then it would be a big announcement for nothing much, because the field of knowledge would be terribly narrow. It's also against the current thinking on the subject, which as Xezlec says involve abstracting words into concepts, and plugging these concepts together to get to the answer. So, in essence, questions like 'What is the largest country?" would trigger the concepts 'geographical area' and 'country', and a humongous taxonomy of everything would return the nodes associated with both, ideally a list of all countries with their size.

Concepts are automatically extracted from a very large database, like, say, a search engine's. They may then be cleaned up manually. But there just isn't any clean database of general information from which to reliably extract the meaning of words. That might be helped by the in-not-so-long-now semantic web, but even after all the information on the Net is tagged, it'll take a couple of decades until we can parse natural language. In the meantime, we can rely on human-organised information, like the CIA factbook, but that severely limits the range of knowledge available.

The second big problem is that the size of the taxonomy to consider gets out of hand really quickly. There is, of course, the astronomical number of different concepts possible. Assuming the engine focuses on a few fields of knowledge only, 'what is the largest military alliance?' would add at least two levels of concepts, but possibly more (country->treaty->active->alliance->military->defensive?). Then there's the ambiguities in language: is largest the most powerful, the one whose countries cumulate the most citizens, the one covering the greatest land area...? Is Madonna a singer, a common Renaissance painting theme, a religious figure, or a metaphor for a respected practitioner of something? How do you extract the right concepts from the database, and how do you extract the right concepts from the query?

I don't think any team in the world can tackle such an astronomically difficult problem just now. There are some serious projects attacking it, but they mostly discover how much harder it is than their most pessimistic expectations.

So, I'm not holding my breath for Wolfram|Alpha. I just really, really hope they've found a clever trick to generically process information within a narrow field of knowledge (so, not an expert system, whose structure would be too specific to the field), because then a bit of mundane work and there's only teleportation left on the checklist.
posted by Spanner Nic at 9:10 PM on March 8, 2009


"Let me Wolfram Alpha that for you." Catchy.
posted by terranova at 9:24 PM on March 8, 2009


Speaking of weird search technologies... whats up with the knowledge generation bureau thing?
posted by gryftir at 9:37 PM on March 8, 2009




I find myself hoping it doesn't work. I want it to produce wrong answers, even inappropriate ones. I hope the answer to every query is "Wolfram's penis." I started using Mathematica back when it was pretty new, and crazy unstable. I had to use it for courses, and I distinctly remember trying to work quickly so I could finish before it crashed. Most problems required three attempts before I could work quickly enough to solve the problem in time. If I saved, it was likely to crash during saving. And this monstrosity has always costed hundreds (I guess now thousands) of dollars. Which came from universities, and students (and me). I can only imagine it's better now, but I have no intention of finding out. This was before he wrote "A New Kind of Science", which I'm convinced was an Andy Kaufman-esque stunt. I don't find him funny. I don't want him to make new software. I want him to fight George Bush, inside an active volcano, and I want Xenu to nuke the winner. [spits in Alpha's face]
posted by Humanzee at 10:07 PM on March 8, 2009 [4 favorites]


My question:

“Who are these people who really want to type out a correct English question to do a search, instead of just banging in a couple of keywords?”

I mean they must exist, since every world-beating new internet search engine for years has been pushing the idea that natural language queries are going to take over. Every failed world-beating new internet search engine.
posted by BobInce at 10:17 PM on March 8, 2009 [3 favorites]


My question:

I left this tuna fish sandwich out on the counter overnight. Is it safe to eat?

OR

I ate a tuna fish sandwich left out on the counter overnight, and now I'm pissing blood. Is this normal, and if it's not, what should I do?

OR

I left a tuna sandwich out on the counter overnight, and then I ate it, and now I'm pissing blood. When I asked my boyfriend what to do, he basically laughed at me, and said he wanted to rent me out to his friends as a "walking tomato juice dispenser." I'm still pissing blood, and I'm also angry and confused. Is this normal behaviour for a boyfriend? How should I followup with him?
posted by KokuRyu at 10:34 PM on March 8, 2009 [1 favorite]


Some sort of searchable database comprising of algorithms from mathworld and the new data they integrated into mathematica 7 (genomes, weather, etc)? All searchable by natural language?

If it works and I could search for an algorithm to do my math for me and process the data it would be pretty cool.
posted by scodger at 10:42 PM on March 8, 2009


Question for the database:

What is the most absurdly obvious conclusion ever reached by a frivolous "scientific" study?

Here's your answer.
posted by ornate insect at 11:28 PM on March 8, 2009 [1 favorite]


Q. "Does the emperor have any clothes on?"

Well, Wolfram Alpha, I'm waiting...
posted by fallingbadgers at 12:13 AM on March 9, 2009 [1 favorite]


I mean they must exist, since every world-beating new internet search engine for years has been pushing the idea that natural language queries are going to take over. Every failed world-beating new internet search engine.

Yeah, that and graphic interfaces, what are people thinking! Text is king!
posted by TwelveTwo at 1:05 AM on March 9, 2009


Also Roger Penrose and Ray Kurtzweil should totally do search engines of their own.

Penrose's would be rubbish, though - it would operate on a quantum level, so the act of asking the question changes the answer.

Kurtzweil's would be fantastically accurate and authorative beyond your wildest dreams, and the response would arrive really quickly. Probably within the next 30 years even.
posted by Sparx at 1:29 AM on March 9, 2009 [3 favorites]


Google makes money because they can throw more and more dumb servers at indexing the web while the engineers work on the core algorithms, optimising for the most common/profitable queries.

Alpha will surely require a great deal more manual 'spoonfeeding' to obtain, update, and structure data sources, while probably leaning towards less profitable search topics/audiences. So I don't hold out an awful lot of hope for it as a viable standalone business, and am always sceptical of anything that hypes 'natural language' features, but it'll be interesting to see what they've built (to either praise or ridicule it).
posted by malevolent at 2:26 AM on March 9, 2009


Lots of illegalities too, if it works:

'What's the GPS location of the nearest hooker?'

'Gimme a breakdown of her rate schedule and menu.'

'Oh yeah, where's the cheapest pot dealer with sensimilla that's on the way to the hooker?'

'Before I leave, which person on the way to the pot dealer has the most money in their wallet AND isn't armed?'

'Finally, where'd I leave my car keys?'
posted by jamstigator at 4:50 AM on March 9, 2009


Besides Lenat's windmill-tiltingest effort, there's stuff like START.

But I think the true test of such a system is for it to tell you what it *doesn't* know, and why.
posted by RobotVoodooPower at 4:55 AM on March 9, 2009 [1 favorite]


Computers are useless, they can only give you answers.
-- The man who was never called an asshole
posted by kcds at 6:22 AM on March 9, 2009


...there's stuff like START.
START's reply

===> who is john galt?

The word GALT may be misspelled.
posted by DU at 7:06 AM on March 9, 2009


I want an AI that answers relationship questions after being trained on Ask MeFi and Dan Savage. :)
posted by jeffburdges at 7:45 AM on March 9, 2009 [1 favorite]


That's the easiest Turing Test to pass ever. You just map all input strings to the response "DTMFA."

Prediction: this DTMFA-Bot provides more useful results than Wolfram|Alpha.
posted by RogerB at 8:06 AM on March 9, 2009


When I think about how much fun it'd be to play around with Mathematica, and then remember how much the damn thing costs.... If this thing works then undoubtedly it'll cost a lot of money for access, and that I hate.
posted by JHarris at 8:06 AM on March 9, 2009


SAGE is as much fun as Mathematica, or more, and it's free.
posted by escabeche at 8:39 AM on March 9, 2009 [1 favorite]


The reason that happens is, assuming the question was formed properly, someone somewhere has formed the response to the question and google finds the exact match.

There is a certain kind of question, like those about bugs in a camera interface, that yield page after page of forums containing the question followed by, "good question, didn't work for me either."
posted by StickyCarpet at 8:57 AM on March 9, 2009 [1 favorite]


From benzenedream's link: So much for substance. Let me turn to the style, which is that of monster raving egomania...

I picked up Wolfram's book last weekend, read the first page, my BS detectors went off, opened to a random page, read it, BS detectors went off. So I grabbed a different book.

The thing that turned me off was the tone of talking down from great heights. Dude, if you're so brilliant, doesn't it seem likely that there would be one or more equally brilliant people in the world, and shouldn't those people be a welcome part of your audience?

To me a true genius, like Wittgenstein, asummes his audience to be both infinitley ignorant and infinitely intelligent, and places himself together with them, showing others how to poke a hole in the paper bag.
posted by StickyCarpet at 9:33 AM on March 9, 2009


StickyCarpet: There are some questions that have not, as yet, been answered. When you get lots of pages of the question and no answers to it, you have found one of these. After a certain amount of time, sufficient for the manufacturer of the product in question to either go out of business or phase out the product in favor of something that works better, these questions are properly to be regarded as philosophical inquiries, and you may attempt to answer them yourself through a process of pure reasoning.
posted by rusty at 9:49 AM on March 9, 2009


escabeche I'll have a look, thanks!

My favorite dismissal of A New Kind of Science was that it should have been named A Fairly Recent Kind of Math.
posted by JHarris at 10:46 AM on March 9, 2009 [2 favorites]


(From this page, it seems.)
posted by JHarris at 10:47 AM on March 9, 2009


Did I leave the baby in the car?
posted by pianomover at 11:00 AM on March 9, 2009


After reading this I still can't understand how this is any different than Cyc... anyone?
posted by Cosine at 12:44 PM on March 9, 2009


Everybody's running around quoting each other, none of them has actually seen the thing but that's not gonna stop em. This is the most precise, least metaphoric explanation I've seen, from Nova Spivak's review:
Wolfram has created a set of building blocks for working with formal knowledge to generate useful computations, and in turn, by putting these computations together you can answer even more sophisticated questions and so on. It’s a system for synthesizing sophisticated computations from simple computations. Of course anyone who understands computer programming will recognize this as the very essence of good software design. But the key is that instead of forcing users to write programs to do this in Mathematica, Wolfram Alpha enables them to simply ask questions in natural language questions and then automatically assembles the programs to compute the answers they need.
It's not much to go on but it's going to be hard to get more than that until somebody breaks the embargo on describing the interface or gives real inputs with real outputs as examples of Alpha's capability.
posted by scalefree at 1:18 PM on March 9, 2009 [1 favorite]


Here's the review everybody's waiting for, the one that actually explains what Alpha does.
posted by scalefree at 5:35 AM on March 12, 2009 [1 favorite]


« Older Mankind up close and personal   |   Flickr [hearts] Panda Newer »


This thread has been archived and is closed to new comments