f13.net

http://bits.blogs.nytimes.com/2009/04/28/wolfram-alpha-veil-lifted/ (http://bits.blogs.nytimes.com/2009/04/28/wolfram-alpha-veil-lifted/)

Anyone gotten a chance to play with this? Sounds really interesting. Would love to know if it actually works as advertised.

From what I understand, basically it's kind of like a search engine but you ask questions and it gives answers. Like if you asked "How high is Mt Everest?" instead of giving links back would just actually tell you how high it is. i tried to read up on it but would love to hear from someone who has actually gotten to play with it.

Stephen Wolfram may be one of the 5 smartest people alive right now, but just about everything that isn't Mathematica that comes out of him or his company right now seems to be a lot of hype with less substance than one would care for.

Conceptually this sounds interesting, but I am skeptical about it, and I tend to give Wolfram a lot more leeway than other people do as I have known quite a few people that work there (and have applied for a number of jobs there over the years) since it is a local company.

I might plunk around with it one of these days.

The Segway of search engines.

I just typed 'How high is mount Everest' into google.


Incredible!

So I used a rather simple example, sue me. But ya the segway comment will probably be appropriate but still, it does sound cool.

So I started throwing oddball "how many/much" questions at Google.

1 mile = 63 360 inches
1 year = 31 556 926 seconds (note this is the number of seconds in an orbital year, not a calendar year)
1 short ton = 907 184.74 grams
1 nautical mile = 185 200 centimeters
1 lightyear = 9.4605284 × 10¹⁷ centimeters

All of these were thrown out by Google's "Calculator" function (http://www.google.com/intl/en/help/features.html#calculator) as the first result in response to a "plain english" query of the default search box.

I found some other interesting things on that "Features" page I hadn't known about, like using a tilde ("~") to search for synonyms of the search phrase in the same query.  Searching for the default "~fast food" example gave me the Wikipedia page for Fast Food and a Google Maps results for fast food restaurants near my house (presumably based on all the many times I've searched for stuff based on my zip code).  I'm wondering why we need Wolfram Alpha.

--Dave

No one likes being made obsolete.

Doesn't really matter, only neckbeards (and I mean true blue neckbeards) would ever use anything named Wolfram Alpha anyway. I mean folks that put ships in bottles and prefer a chalkboard to a whiteboard.

Yeah asking Google how to fix a prolapsed anus turned out plenty of results.

What, I needed something more obscure than you people and your silly common knowledge.

On the other hand asking "what is the maximum span of a 2x10 syp#1 joist @ 19.24"O.C." returned less useful information.

Cutting that query down to a more reasonable "what is the maximum span of a 2x10 joist" returned this calculator (http://www.awc.org/calculators/span/calc/timbercalcstyle.asp) as the first result, which seems to have enough options to meet any particular need you might have.

I'm remembering what it was like to search pre-Google, remember how rare it was to find something useful in the first page of results from AltaVista even before the spam links got started?

--Dave

I lol'd.

Well google failed pretty badly with some science questions, so I'll wait and see how Wolfram Alpha does with those.

"Do unicorns exist? Addendum: If not, why is there a horse with a gold horn in my backyard?" isn't a science question, K9.

I went to a talk 5 or 6 years ago by this man: http://en.wikipedia.org/wiki/Lotfi_Zadeh

We've been conditioned into thinking like the search engine when we ask a question to get decent results but there is A LOT more search engines could do that they don't.

Try something like, "Name a few large lakes in Switzerland that are near France?"

It's not a hard question, it would take a human about a minute with a map to come up with a pretty good answer with a small set of candidates and is an example of a style of question, that if it could be answered by a computer would be very useful.

Being able to interpret fuzzy concepts that relate two different things is really where searches need to be headed.

More here: http://www-bisc.cs.berkeley.edu/zadeh/papers/From%20Search%20Engines%20to%20Question-Answering%20Systems...2006.pdf

48 pages  :|. Time to buy a kindle.

As a librarian, the possibilities for this technology are simply awesome. 

It puts me one step closer to building that bunker in my backyard though, so when the machines take over I'll join the resistance to fight back. 

Wake me when it's live and available for public consumption.

Otherwise, it's just PR puffery.

It went live and it seems to work pretty decently. 

I asked it "What is the answer to life, the universe and everything?" and it answered "42". 

I'm down with that answer. 

OF COURSE YOU DON'T, MR. ALPHA.

What the fuck is this thing?

The thing on the right says, "Enter any food" and then gives an example of "1 apple + 2 oranges"

Ok.

So I enter "1 apple + 2 oranges"


Uh. Ok. Maybe we should stick to using it as a calculator and use, oh, I don't know, the rest of the internet for nonsense.



However! It gave me a unit of measurement for mudkips, which is kips. Obviously.

it's scaring the GOOG which is good enough for me.  Big song and dance this week from them about the 130-some products they've launched and how they are S-M-R-T about serch.

"What is the circumfrance of the Earth?"  --  Dunno what to do; link to earth which has radius but not circumfrance

"What is the mass of uranium?" --  Got it plus conversions

"Is there a chemistry department at UH?"  --  Dunno

"University of Houston Chemistry Department"  --  Dunno; link to UH which gives UH stats, but nothing about my department

"How long would it take to reach proxima centauri?"  --  Dunno; link to Proxima Centauri with lots of info

"What is a black hole?"  --  Gave some neat equations.

/yawn

Lethargic response time, most queries unable to "compute", offering up basic information that you could already get from Google (you are aware you can use Google as a calculator + dictionary), and using CIA factbook or inserting "wiki" at the lead of your search query will net you most of the stuff they're providing without the flash progress bar fanfare…

/wake me when it can actually do something remarkable…

So, when you put in something it doesn't understand, for example "u" instead of "you," the engine adds it to a list of future topics. Perhaps something it or the people behind it will research.

It gives me an interesting idea. Learning computers aren't going to take over the world if we make them stupid.

Perhaps I really want to know if "it lieks mudkips" now.

It doesn't seem to respond well to phrases.

e.g. "What does the NOD2 gene encode?" gets the standard "Wolfram|Alpha isn't sure what to do with your input." However, simply entering NOD2 gives you gene information. Same result for FOXP2.

I'm trying some more ambiguous acronyms though, since gene nameses are more likely to be distinct, and the engine seems geared towards sciences.

Entering "BBC" gives: "Assuming "BBC" is an airport" and gives information about Bay City Municipal Airport, TX. The "referring to a general topic" link does point to the Television and Radio channel, but nothing there about the BBC. Wikipedia already resolves conflicting acronyms and entries far better than this.

Entering "MMA" gives: "Assuming "MMA" is a medical test" and then some brief notes about some medical procedure.

Simple phrases for simple questions gets some results:

"What is tea?" -- Dictionary definition of tea "assuming tea is a word"

"Who is schild?" -- Decides that schild actually means schilde and gives you information about a town in Belgium

"Who won the Battle of Gettysburg?" -- "Wolfram|Alpha isn't sure what to do with your input."

"Battle of Gettysburg?" -- Gives you the distance between Battle (a small town in the UK) and Gettysburg in the US.

"e^(pi*i)" -- Gave correct information about this no problem.

"Nonlinear optimisation" -- "Wolfram|Alpha isn't sure what to do with your input."

"Nonlinear optimization" -- "Wolfram|Alpha isn't sure what to do with your input."

"Methicillin" -- "Wolfram|Alpha isn't sure what to do with your input."

"Swine Flu" -- Gives case and death statistics for the last 5 days, nothing else


I have to agree with naum, nobody is going to find W|A more useful than a combination of google and wikipedia anytime soon. Even when you do get an answer that relates to what you intended to ask, the detail is very thin; and of the detail it does give, most is useless. Searching wikipedia for "Swine Flu" will give you far more information than this; the choice of information which W|A is giving you seems bizarre in comparison.

Enter "You're not perfect" and I get the response

(http://www.filedump.net/dumped/wa1242495022.JPG)

Probably coincidental, but paused me for a moment when I read that.

Alpha is pretty neat when you hit something it knows about, but pretty sad when it misses.

Compare "cats" on Alpha and Google -- neat taxonomy dealie vs wikipedia page and assorted other resources on cats.

Compare "yellow boxfish" on Alpha and Google -- nothing vs some nice pictures, and wikipedia as the second hit.

Compare "DS2431" on Alpha and Google -- nothing vs the manufacturer's page datasheet.

EDIT: Reading more about it, guess it's a "knowledge base" and not a search engine.  Hopefully it'll start turning up in google results and I'll get the best of both worlds...

What is the key of A? Some crap about amperes.

What is a major key? Some crap about ethnicity.

What is the scale length of a Gibson SG? Nothing.

Estimated average cruising airspeed of an unladen African swallow? Answer given.

Geek toy.

http://forums.f13.net/index.php?topic=16870.msg645171#msg645171

Result:
there is unfortunately insufficient data to estimate the velocity of an African swallow (even if you specified which of the 47 species of swallow found in Africa you meant)

Also:

Input interpretation:
estimated average cruising airspeed of an unladen European swallow
Result:
25 mph (miles per hour)\n(asked of a general swallow (but not answered) in Monty Python\'s Holy Grail)
Unit conversions:
40 km/h (kilometers per hour)
21 knots
11 m/s (meters per second)
660 m/min (meters per minute)
0.66 km/min (kilometers per minute)
950 km/day (kilometers per day)
950 000 m/day (meters per day)
0.41 mi/min (miles per minute)
590 miles per day
36 ft/s (feet per second)
Corresponding quantity:
Slowness from S=1/v:\n | 0.091 s/m (seconds per meter)

So far where it seems to shine is for a search that you want it to return information on a top that would give to many results on google.

I don't know much about the architecture of this thing.  Is it learning?  Might we expect it to ramp up in knowledge as it gets used more often?

It's a warehouse full of minimum wage people inputting answers from google into a keyword database. So Schild asks about swallows, no answer. Later that day some nerd types an answer in for Sky to see. No emergent algorithm, google and a finite number of monkeys.this may be speculation

Nah, seems accurate. Once someone types in enough things, it gets added to future topics. It's not really learning so much as people who are probably interns who get to work "some place smart" doing something they believe to be noble in the pursuit of knowledge or some shit.

Try getting Wolfram Alpha to tell you who won the last F1 motor race.


I'm using "F" for something else now. F Wolfram Alpha.

It's not porn compliant. I predict failure outside the academic community.  :awesome_for_real:

Give 4chan a chance, it's only been live for a few days. 

Their terms of use are pretty amusing, can't use their answers to anything without attribution, as they claim copyright on all search results.

So pretty much, fuck 'em.

 :oh_i_see:

Loaded in a number of questions to do with philosophy and it wasn't sure what to do with any of them, I have a strong feeling this is going to be more useful for scientists as far as academic research goes. Even just plugging in a load of keywords got me nothing while Google has reams of papers and books and random websites.

They're trying to patent the internet. oO

It is going to struggle to offer anything to the medical and life science that isn't already offered in a way that is far more tailored to the needs of the end user; and it has to be more than just 'Mathematica online' to get anywhere. As I understood it was supposed to be equivalent to AskJeeves, but with the ability to actually handle questions phrased as humans speak, something which it completely fails out.

The majority of scientists are quite web-literate don't need a dumbed down interface to access the resources they need on the internet, and furthermore tend to ask the most complex questions. So they don't seem to fit as a demographic that would actually use W|A as anything more than a toy.

I actually found a use for it this evening.  The wife was looking at a little article about the Bugatti Veyron where it said it goes 0 to 60 in 2.5 seconds and she said 'wow, I wonder what that would feel like' and I said 'I dunno.  Let's figure out what it is in Gs'

So I inputted:

And:
Fuck yeah!  Now I understand the VISION

(http://dl-client.getdropbox.com/u/39720/f13/1.09g.png)

Took about 0 seconds.

onebox to the rescue! (that's the google lingo for the special features that recognize things like calculator input for special handling):
http://www.google.com/intl/en/help/features.html

http://www38.wolframalpha.com/input/?i=how+many+roads+must+a+man+walk+down

I'm down with that :)

I swear W|A has more pop-culture and geek answers right now than actual answers  :oh_i_see:

Oh noes!  My Wolfram Alpha phase is over.

Edit:  I should also point out that in my original WA effort I wanted to figure out how that compared to the G forces in taking off in the space shuttle.  And all kinds of queries like 'Shuttle launch force in gravs', 'Space shuttle acceleration', etc, etc were all no worky.  Google gets it on the first hit.

NaN

That's also just about the ragged edge for soft rubber on concrete traction, so we're not going to see anything much higher until they invent a stickier material to make tires out of, or start covering test tracks with adhesive.

--Dave

Just for shits and giggles, I tried the question "Where is a good Thai restaurant in Austin TX?"

Wolfram Alpha: isn't sure what to do with your input.

Google: First result was Local business results for Where is a good Thai restaurant in Austin TX (http://maps.google.com/maps?hl=en&um=1&ie=UTF-8&q=Where+is+a+good+thai+restaurant+in+Austin%3F&fb=1&split=1&gl=us&view=text&ei=GsgUStrwJ4-CmQfd-oHmAw&sa=X&oi=local_group&resnum=1&ct=more-results&cd=1)

Clicking on that took me to a page where I could limit by Review score of 4 stars and higher, and it would take some more fiddling to get one sorted by how close they were to me, so there's room for improvement (especially for Android phones that won't have to be told where I am).  But if I'm looking for a Thai restaurant in a strange town that won't make me gag, I think Google will be much more useful than Wolfram Alpha for quite some time.

And unlike the iPhone, I won't have to pay money for an app that does nothing but pre-filter a search result.

--Dave

It's true, all W|A has done is made me appreciate Google more.

I guess this is what happens when you put a bunch of scientists together in a locked room instead of a bunch of interesting fuckers.

I've found WA is handy for quickly graphing and assessing the properties of multivariate equations and finding if factors of polynomials exist. Also, Google fails to assess the health benefits of a cubic lightyear of strawberries (http://www72.wolframalpha.com/input/?i=cubic+lightyear+strawberries)

F1 cars do 1.45g using rubber tyres on normal asphalt.

6.3 x 10⁵³% of your daily value of Vitamin C is admittedly a lot, but you actually just piss out all of the excess anyway.  Sadly, WA is not informative in regards to the properties of a cubic lightyear of urine.

I'm pretty sure a cubic lightyear of strawberries will collapse and form a super massive black hole.

Yeah, you've gotta eat them before they go bad.

Going bad is worse.  Then you've got a super massive black mold.

And need a new set every hundred miles or so.  But point made.

--Dave

Wolfram|Alpha isn't sure what to do with your input.Tips for good results »

Related inputs to try:

None
    * cubic lightyear

Food:
    * nutritional information
       bacon

/shrug

However, I'm pretty sure I could enjoy a cubic lightyear of maple syrup (http://www72.wolframalpha.com/input/?i=cubic+lightyear+maple+syrup). Mostly, I want to go swimming in it. Maybe I'll drink a little. GLUB GLUB.

Heh (http://www.google.com/trends?q=wolfram+alpha)

That was automatically compiled, wasn't it?

Because being teabagged by a competitor's data collating algorithm is awesome, and the prime reason why you just don't fuck with Google.