Wolfram Alpha: ‘People Just Need What We Are Doing’ | Hacker News Reader

Wolfram Alpha: ‘People Just Need What We Are Doing’

wired.com

71 pointsspottiness45 comments

I tend to wolfram it when googling it fails. E.g. when considering buying a new place in Tokyo:

http://www.wolframalpha.com/input/?i=historical+interest+rat...

Compare those awesome results to the crap google offers:

http://www.google.com/search?safari&rls=en&q=histori...

Google is more generally useful, and has changed all of our lives for the better. Nevertheless, Wolfram Alpha is far more useful in certain contexts, in my experience, and not necessarily just for mathy stuff.

http://www.wolframalpha.com/input/?i=savings+rate+US vs http://www.google.com/search?q=savings+rate+US

no contest.

I use wolfram exclusively for all my economics homeworks. Google is quite useless for that.

Say I'm crunching mortgage data. The engine needs to hit

http://www.federalreserve.gov/econresdata/releases/mortoutst...

and calculate totals & graph that. Wolfram does those things. Google doesn't.

I like Wolfram Alpha, but despite being way towards the top of the "people most likely to find a use for Wolfram Alpha" list, I've only found a use for Wolfram Alpha a couple of times. I use Google Calculator all the freaking time, and only resort to Wolfram Alpha on the odd occasion that Google Calculator fails me.

Does anyone use it regularly?

I use it for unit conversions sometimes, but its parser and its understanding of Gaussian electromagnetic units leave a lot to be desired.

It's also good for a quick Taylor series when I don't feel like firing up Sage.

I used it to create the graphs and compute the solutions for my recent blog post:

http://www.solipsys.co.uk/new/AnOddityInTennis.html?HN

I found it really useful to have a facility on-line that let me do those calculations and draw those graphs.

I'm interested - what would you use instead?

But that's not an on-line solution. I have half a dozen desktop solutions. I don't (yet) use R, although it's on my "To Do" list for my Copious Free Time(tm)

Yes, I use it regularly. Most common is for the kinds of things you might find in the CIA Factbook, or any number you'd expect to be buried in a Wikipedia page.

The most recent thing I was impressed by was this: http://www.wolframalpha.com/input/?i=time+to+sunburn

I also tried creating a widget with WolframAlpha that seemed potentially useful: http://beeminder.com/agename.html (ie, getting a probability distribution on a person's age based on their first name)

It says my name is not a valid input, that's pretty harsh.

So did Renren, the China-based Facebook clone.

Characters in the Latin alphabet do not count as names when one tries to sign up there :(

The sunburn one is a cool concept, but most of the data I put in tells me that sunburn is unlikely in Phoenix, AZ.

Or that a sunburn is unlikely in Miami, Florida (which is where I am).

That's because it was night time. :) You can actually include a time and date.

Another handy heuristic: worry about sunburn when your shadow's shorter than you.

I use it for pretty much all calculations, solving and rewriting equations, currency conversions, and basic stock comparisons. It's way better than google calculator for those things. Just add it to your bookmarks bar.

I use it for conversions and as the go to calculator on my iPad.

When I was moving/job hunting I used it a lot for cost of living adjusted comparison of salaries.

All the time - use it to check my answers to pretty much every calculus question I do, then if I get any wrong just use the Steps to work out better approaches in future. Never really looked into Google Calc though.

Not regularly but I used it recently at work to compute symbolically the sum of a non-trivial series. Turned out the sum has a closed form, which allowed me to replace an O(n) for loop with an O(1) bunch of additions, multiplications and divisions. Felt nice.

Can you give an example?

Sure, here's one: given a list of n items, compute the sum of pairwise distances between every pair of items.

The first item has distance 1 from the second, 2 from the third, ..., n-1 from the last. Similarly, the second item has distance 1 from the third, 2 from the fourth, ..., n-1 from the last. And so on until the (n-1)-th item.

Expressed as a finite sum, this amounts to:

  sum (sum j, j=1 to n-i), i=1 to n-1

Paste this to Wolphram Alpha and you get the answer in a closed form: http://www.wolframalpha.com/input/?i=sum+%28sum+j%2C+j%3D1+t...

I use it for time calculations (timezones, differences, etc.)

I'd say that this is where Wolfram falls down the most: even those of us who /probably/ are smack dab in the middle of the farther audience don't really understand the use cases. If Wolfram can figure that out, or (more specifically) if I could figure it out for me, then I'd use it much more often. But until I get that handle on it...

I do. I like to think and walk. Being able to whip up my phone and say something like plot d/dxdydz of [some equation in 3 terms] or maximize f(x) or solve some differential equation in a mix of english and symbols is very convenient.

On chrome, I also use https://chrome.google.com/webstore/detail/acgimceffoceigocab....

What wolfram alpha is not good for is the trivia natural language stuff that they touted. Its very hit and miss, awesome when it works but often doesn't. Stuff like trying to find a list of countries orderable by gdp,average income, democratic index, crime is easier to put together using google. Its not just about comprehension, a lot of the data just is not there. A query that I was surprised it didn't get was "basketball player with highest field goal percentage". google got that easy.

I use it regularly for nutrition (calorie counting) and comparing foods, e.g. 100g of salmon, 100g chocolate etc. http://www.wolframalpha.com/input/?i=100g+salmon+100g+chocol..., or http://www.wolframalpha.com/input/?i=potassium+in+1+cubic+li....

Yeah, love the google calculator. I wished it had support for high precision numbers.

Most people who've commented about using Wolfram Alpha are just using just its Mathematica backend - calculations, plotting graphs etc. I found this pretty useful while trying to solve some algorithmic coding challenges which involved large numbers. Mathematica works with numbers larger than what either Python or Ruby can support.

But otherwise apart from the Math part, haven't heard any use case yet where the data analysis that Wolfram Alpha does was terribly useful.

Python integers have unlimited size. It’s true though that the support for fractions, numbers with whole multiples of π and e, square roots of arbitrary numbers, etc., is better in Mathematica/Maple than a general-purpose programming language.

I find it very useful when dealing with unit conversions and fundamental constants. Google is fine for simple stuff but Alpha really excels when you have extremely complicated combinations of units. This probably doesn't matter much outside of physics but within it Wolfram Alpha is a godsend.

When I'm planning out my meals I use W|A to calculate calories. It's incredibly useful for this. I love that it not only gives point estimates for fat, sodium, etc., but also plots the distribution of these values: http://www.wolframalpha.com/input/?i=4oz+turkey+breast Other than this, very special case, I never use it. And I am a former Mathematica fanatic.

"I’ve been lucky enough to run a company that’s been profitable for 23 years, so I developed the habit of doing things that way [..] That’s a way of doing business, that if you think about it, computes much better than getting tens of millions in funding for an iPhone app."

But Wolfram is frustrated a bit that users don’t know the full power of Wolfram Alpha. “The mental model for when to go to Wolfram Alpha is not fully fleshed out yet,” Wolfram says.

This frustration reminds me of Google Wave, a wonderful idea, on the verge of being totally disruptive, but that the average user wouldn't grasp. Hopefully WA can sort that out, otherwise it may just remain a web API to a data-rich Mathematica, which in any case is a nice thing to have.

I purchased the Android app, but I find I just don't use it that much. I'm not even sure that it's that the 'mental model' isn't fleshed out yet, it's just that Google got there first. Search, scan, refine. It's how I think now. Maybe there is a space for Alpha at the end of that sequence.

Still waiting on W|A to incorporate basic usability improvements I proposed back in 2009:

http://codingthriller.blogspot.com/2009/05/fixing-wolframalp...

Queries that I have are usually answered by DuckDuckGo/Wikipedia.

Running a query on 'linux users' returns, among other things, phone keypad digits for 'l-i-n-u-x', and Scrabble scoring.

Not what I need.

It's not a search engine. It's an information engine. Enter what you want, not just two random words with no 'question'.

I then asked 'how many people are using Linux?'. It seems to think I'm asking 'how many people use the word 'Linux', as it returns the same result set as my initial query (word definitions and such). I ran the same query again with 'Linux' as an operating system topic, and it tells me that "Development of this topic is under investigation.."

Then it asks me to leave my email address to show interest.

And I would point out that the same query, given to Google, returns a suggestion to search for 'number or Linux users', which is what I was thinking. Google doesn't know much, but it knows enough to get me where I'm going without having to think, or type, much. Isn't that what it's all about?

I understand the geeks downvoting me, though. Wolfram Alpha is 'really neat' and 'really neat' is what we like.

What I don't understand is this idea that hammering a nail with a toolbox is a smart way to do things.

Try these in Google:

    flights visible from Modesto, CA
    cepheids visible from Modesto, CA
    distance from Mars to Jupiter

Now try them in WolframAlpha. See the difference?

Google is a search engine. It tries to find web pages that it thinks are relevant to those searches. It is very sophisticated at this, so as to try to figure out what pages actually have the information, and which just happen to have the words. However, since these are things that probably no one has written about on the web, you get nothing really useful.

WolframAlphas is NOT a search engine. It treats those same queries as requests to try to answer them from its databases or by computing them from things that are in its databases. So, for the first it figures out that you are interested in scheduled airplane flights that you can see from a particular location. It has airline flight data and radar tracking data, and from that, and it's model of visibility, it figures out what planes could possibly be seen from Modesto, taking into account their last known position and heading and extrapolating to account for the fact that the radar data it gets is behind by something like 15 minutes. (Click on a flight in the resulting list, and it gives you neat information about it, such as when it took off, how late that was, a plot of its course, graphs of altitude vs. time, groundspeed vs. time, and heading vs. time).

The second is similar. It figures out that Cepheids are a kind of star, gets a list of the 100 brightest Cepheids, and figures out which are above the horizon at Modesto, CA, right now.

For the third, it decides you are talking about planets (although you can tell it you mean Jupiter, FL and one of the several cities named Mars if you wish), and from their orbital information (which it has in its database) it calculates where they are now, and tells you the distance between them.

Again, that's all 'really neat', but the headline of the story is stating that people need what they're doing, which is b.s., so long as I can't ask how many people are using Linux and get something like an answer. I don't need to know how many cephids are visible from Modesto, CA. If you do, Wolfram Alpha is 'really neat'.

The beautiful thing about Google is that most questions along the lines of 'what is the distance between Mars and Jupiter", or "how many mililitres are in a gallon?" have already been answered, somewhere, and Google will point you to the answer. If it aint broke, don't fix it. Now, I don't doubt there are applications for Wolfram Alpha (I've heard their API is what really gets people excited), but do 'people need what they're doing'? Not particularly. It's a lot of hype.

Stephen Wolfram is sort of famous for his ego. That explains the title.

The first result on Google for "what is the distance between Mars and Jupiter" is pretty close.. today. It's off by 100% if you want the distance on my birthday.

Many problems don't have static solutions which can be neatly cached by a search engine.

It sounds like you're just tryin to be argumentative. People need insulin shots too, but it would be rather pointless for a non-diabetic person to call B.S. on that. If you ever go to a funeral and you hear someone say "Everybody liked him", please do the bereaved a favor and don't correct the technical flaw in their statement.

people != all people

Note that wolfram alpha can also help you with logic.

Now try something useful like: "where to buy low cost computers in miami florida"; or "what is the temperature of the water off the coast of miami beach".

Now try it in Google. See the difference. ;)

You right, WolfmanAlpha is not a search engine. It's a fancy encyclopedia.

Reply on news.ycombinator.com