I just don't get it

[andrewducker]

There are some things I am yet to understand the use of.  They get a lot of hype, but they don't seem to do anything I can see myself actually wanting.  Frequently I can't see their use to anyone else either.

3D user interfaces are one of these - every so often someone shows off something that looks incredibly cool, but is clearly about 10% as usable as, say, the standard Windows/Mac/Gnome interface.  They promise to take their new wonderful interface and make it more useful, and that's the last anyone ever hears from them.

And the latest thing to fit into this category of "sounds shiny, can't see the point" is Wolfram Alpha.

Can anyone explain what I might actually want to use it for?  Because I've played with it a couple of times and I can't figure it out.

Comments

[kashandara.livejournal.com]

I realise it is partly my own fault for being stuck using IE6 (work computer, no way of upgrading/changing) but the webpage completely fails to load in any sort of useful way for me so I doubt I'll be finding it useful or shiny anytime soon =)

[nancylebov.livejournal.com]

It doesn't load for me, either, and I use firefox. It loads for other people. Maybe they're using Opera.

[andrewducker]

I'm using Firefox.

[drplokta]

It's good for specific financial stats -- try entering "uk gdp in march 1996" into Wolfram Alpha and Google, and see which is more useful.

[andrewducker]

Interesting. I played a little more, and got this - which looks unlikely:
http://www58.wolframalpha.com/input/?i=eu%2Fus+gdp

[sigmonster.livejournal.com]

Looks about right to me, since the EU is a larger economy than the US after the most recent round of expansion. (Per capita gdp is lower in the EU, though.) Remember that Italy, France, and the UK are all $2 trillion economies, Germany is $3 trillion, and there are another 23 member states to count.

Of course exchange rates can change very rapidly...

[andrewducker]

I get a nice smooth graph showing that the EU used to have 16x the GDP of the US, but now only has 1.2x

That change seems unlikely...

[sigmonster.livejournal.com]

So it does, sorry!

It seems to be using the constant 2009 value for the EU and a time series for the US (US gdp in 1970 was a hair over $1 trillion). But if it's smart enough to work out that an EU time series is ambiguous (all current members, or just membership at the relevant date?) then it should be smart enough to annotate the output, too...

[drdoug.livejournal.com]

I'm waiting for the Beta at the earliest.

[channelpenguin.livejournal.com]

ROFL!!!

And I keep thinking of Alasdair Reynolds type Alpha simulations...

[bracknellexile.livejournal.com]

I suspect it's very much a proof-of-concept thing at the moment. Once its available datasets vastly increase (and they refine the semantics engine) it might be a useful tool. For now there are massive gaps in what it understands but then it is only an alpha.

It may go the way of Cuil, it may become the definitive companion to Google when looking for quantitative rather than qualitative results. TBH, I suspect if anything does come of it, then it'll be Wikipedia losing hits to this more than Google.

[rosathome.livejournal.com]

It said 'Enter your question or calculation' so I asked it 'How long is a piece of string?' Answer: 'Wolfram|Alpha isn't sure what to do with your input.'

Hmm.

[robhu.livejournal.com]

But where else would one compute how many calories there are in a cubic light year of ice cream? (notice how you can then refine your answer by specifying the ice cream flavour)

Also, The Register's review is amusing.

Still - it is kinda fun, is useful for doing some maths stuff if you can figure out the syntax (and don't have Mathematica to hand), and when the data sets improve / the language parser gets better, it might actually be useful for that too.

[kurosau.livejournal.com]

Is that a comedy review, because man, what a shitty review if it's serious.

[broin.livejournal.com]

With the Reg, it's always a little from column A, a little from column B.

[the-magician.livejournal.com]

I can already imagine a 3D user interface and how it would be better for me, but I don't know anyone that is developing it.

It's not much like the 3D skyscraper layout from Jurassic Park, but would involve something like the distance fades for older documents so that things at the front (or "top of the desktop" are the most recent (if you've ordered it that way) and stacked by categories (so roughly like folders) and you could zoom in/down/backwards in time, using other calendar and document based cues to find a document that you did, say, last spring (have the calendar pages fall up and back onto the calendar, have the snow melt, autumn, then summer then last spring on the trees out of the window, etc.) and find the document you created then.

My *real* desktop is an exercise in 3D and chronological filing, as are things like noticeboards at home/at work ... use the 3D paradigm to bring that onto the work desktop.

For phone numbers up near the phone at home, some are in different colours, some are scrawled sideways, some are business cards ... and my mental filing system knows that, say, Bob's number is in red felt-tip in the top right hand corner so I can look there instantly when I want it ... and that the cycle map of cambridge is under the phone books in the hall ... see The Art of Memory for how these visual cues of associating a place with a fact have been used since at least Roman times as a way of learning, organising and retrieving information.

Turning it into a useful 3D workspace will make someone rich/famous!

I tried out Wolfram Alpha, it produced no useful results on any query I gave it, from my name, to the membership numbers for Eastercons. It either had no idea what to do with my input, or gave me a random result that sort of included one or more words from my query. My needs and what it can provide appear to be a disjoint set at the moment.

[kynn.livejournal.com]

I used that 3d interface from Jurassic Park ... and boy was it awful.

[khbrown.livejournal.com]

But it does let you go "It's a UNIX system! I know this!" ;-)

[bracknellexile.livejournal.com]

I think the big problem with Wolfram Alpha at the moment is that people are expecting an all-singing, all-dancing all-knowing repository of information.

As the Register's article points out: The number of data sets included are, at present, limited to a few hundred, although, as a spokesman for Wolfram pointed out, as one data set includes "all current and historical weather", whilst another includes "the English language", this measure may be deceptive.

Since the results are returned from computations within Wolfram's servers and not directly from the whole net, ala Google, it's not that surprising that so many queries produce no result at the moment. If the semantics engine really can refine the way we ask the net for information then it's simply a matter of adding data until it's useful. As the Reg also points out, enter "gold" and "lead" and it'll return chemical element info. Enter "gold" and "red" and you get colour information. Enter "Cambridge" from a UK IP address and you're pointed to the city in East Anglia. From a US IP it goes to Cambridge, Mass. Google doesn't have that level of interpretation. Even at this early stage, it's that that makes it something different - it just hasn't got the volumes of data to draw on yet. So yes, peoples' needs and what it can provide are mostly a disjoint set right now, but I think it's got great potential. It is only an alpha after all.

[the-magician.livejournal.com]

Enter "Cambridge" from a UK IP address and you're pointed to the city in East Anglia. From a US IP it goes to Cambridge, Mass. Google doesn't have that level of interpretation.

HA HA HA HA ... *wipes tear from eye*

I work walking distance from Cambridge UK city centre, my PC has a US IP address, *every* Cambridge search I do in Google gives me US answers first.

Of course if I narrow it down to a street name that only exists in Cambridge UK, or add a postcode, then Google gives me local responses, but if I ask for "Cambridge Bicycle" I get a bunch of Cambridge and Boston Bicycle shops, bike routes etc.

[kurosau.livejournal.com]

Isn't Wolfram Alpha the search engine that's eventually going to be able to understand what you mean when you say "Who was that guy that was the butler on that TV show starring Valerie Bertinell"?

[andrewhickey.livejournal.com]

It's not just you. Something like that may be a useful data-mining tool if it were ever done properly, but at the moment it's just pathetic, and more a monument to Wolfram's ego than anything else...

And the first thing Holly did when she installed Ubuntu was turn off the 3D effects, so she could actually get stuff done...

BTW Google now want to have a '3D web'. God help us all...

[andrewducker]

Oh, I don't mind shiny transparency effects - they don't slow down my laptop at all. Actual 3D interfaces are a different matter.

And 3D stuff hosted in web pages sounds like a great idea to me. Not to replace the web, but to add to the types of objects you can get to me without having to additional install software.

[aliiis.livejournal.com]

Oooh, tell you what can do which isn't actually useful per se but I thought it was cool - is show you quite interesting data on the popularity of given names, although only for US population as far as I can see. This is if you are as unimaginative as me and the first thing you do is type 'alice' into it.
Check it out, looks like people called Andrew are much younger than people called Alice. That means you and I are bucking a statistical trend! (yes I know, kidding, KIDDING)

[andrewducker]

Actually - nice graph here:
http://www99.wolframalpha.com/input/?i=andrew+alice

[aliiis.livejournal.com]

Huh, yeah! That's pretty neat I think, but would be better if you could compare different countries, or continents or something.
This is silly though:
http://www99.wolframalpha.com/input/?i=alice+ben

Which obviously led me to try:
http://www99.wolframalpha.com/input/?i=predator+commando
I... I sort of wish there was a graph for frequency analysis of Arnie/Bill Duke/terrible post-mortem one-liners, you know? Needs moar graphs!

[0olong.livejournal.com]

OK, it may be almost completely useless now*, but SOON it will know EVERYTHING IN THE WORLD and understand it when you ask about it!

Or, less sarcastically, it seems like a fine idea, but appears to be let down by the same problems that dogged 'A New Kind of Science' so badly - bombast, self-importance, over-reaching and militant unawareness of its surroundings...

* I just found a use! Nutritional value of 1 peanut. This information was surprisingly difficult to pin down before. I forget why I was trying. Down side of this find: It didn't understand when I asked for 'nutritional value of 1 peanut', or 'nutritional facts about 1 peanut', or 'nutritional information on 1 peanut'; it only understood when the request was simply '1 peanut'.