Ask HN: Whatever happened to Wolfram Alpha?

I use it regularly. Sometimes it’s broken, and maybe nobody notices but me? :)

Their natural language queries for things that I know they know about are amazing. Here are some that I have used recently. You really need to see these results to appreciate them.

I wanted to know how tall my daughter might be.

   8 year old female 55 lbs

http://www.wolframalpha.com/input/?i=8%20year%20old%20female...

I wanted to know the nutrition content of an egg sandwich.

   1 egg, two slices whole wheat bread, one slice of cheddar, two pieces of bacon

http://www.wolframalpha.com/input/?i=1%20egg%2C%20two%20slic...

I was curious about the relative usage of two names over time.

   Michael, Henry

http://www.wolframalpha.com/input/?i=Michael%2C%20Henry

Also a frequent WA user. I use it for things I could calculate, but are much faster to just ask in plain text.

How much that cloud instance really costs

  $0.03/hr * 1 month

Bandwidth calculations for hosting providers

  10 TB per month in Mbps

Yeah, it's great for these types of things. It also has a bunch of values built in, so you can do things like:

  (day length of jupiter) * 80

Yeah; I use it for the occasional repeating specialized query, but have never broadened my usage to anything more-general.

The sandwich example was brilliant! I never expected that to be possible (the example of packing smaller circles in a larger one in another comment is also brilliant but less useful for me today I think.)

1. It's slow, even for simple microsecond computations like log(2). Takes about 5-20 seconds to load a page on my 1Gb fiber connection. Opening Python/SymPy Gamma is much faster for most things. https://gamma.sympy.org/input/?i=log%282%29

2. Every time I use it, a box saying

    NEW: Use textbook math notation to enter your math. TRY IT

pops up over the result, and clicking the X doesn't hide it the next time I search. This adds ~3 seconds to the result time.

3. I'm a long-term Mathematica user, but typing literal Mathematica syntax usually never works, except for simple expressions.

4. Results are PNGs, and copy-pasting a numerical result takes a few unnecessary clicks. "Plain Text" > Copy.

> Opening Python/SymPy Gamma is much faster for most things.

Is there a way to make it plot multivariate functions? I tried but whenever I enter two variables it says "Cannot plot multivariate function." I've seen many Python packages plotting multivariate functions so I'm convinced it should be possible.

I don't think so. You'd need to run it in a terminal with something like

    from sympy.plotting import plot3d
    x,y=symbols('x y')
    plot3d(x*y, (x, -10,10), (y, -10,10))

> Takes about 5-20 seconds to load a page on my 1Gb fiber connection

Wolfram Alpha is implemented in Mathematica, which --- to understate the situation --- was never intended as a high performance backend server language. I suspect that's the reason for the bad performance.

"As a result, the five million lines of Mathematica code that make up Wolfram|Alpha are equivalent to many tens of millions of lines of code in a lower-level language like C, Java, or Python." [1]

Sure, there's something to be said for implementing logic in high-level code, but without a plan for lowering that high-level logic to machine code in a way that performs well, you're setting yourself up for long-term pain.

[1] https://blog.wolframalpha.com/2009/05/01/the-secret-behind-t...

I doubt the bad performance is due to evaluating expressions itself. If I type N[Log[2]] into Mathematica, it evaluates in less than a millisecond. It's probably because Wolfram Alpha is using natural language process to try to process my query and then finally deciding that by N[Log[2]], I mean N[Log[2]]. And it's probably not because of that, but because their grid scheduler isn't optimized for sub-second latency.

Ha, hearing the word "process" in Wolfram's voice, there.

Your Internet bandwidth is not relevant when talking about a compute-heavy backend like this. Wolfram|Alpha is not going to load any faster on a 1Gbps connection than it will on a 20Mbps connection, other than some static assets, but even that isn't going to be hugely noticeable if we're talking about 2ms RTT on fibre vs 8-20ms RTT on cable/DSL. If you're downloading a giant file off a nearby CDN, then sure, 1Gbps fibre is useful. I can max out my 1400Mbps cable connection downloading things this way (it's mind-blowing...), and my latency to my upstream gateway outside of my house is 8ms. But Wolfram|Alpha isn't going to load 40% faster for me than it will for you since it's I/O bound and your end-to-end latency is waiting for the backend to complete your request.

I will say, though, that Wolfram|Alpha could be "optimised" in the sense that it could do less fancy JS and be a simple box with a submit button, like SymPy Gamma.

I think that's the point. "My internet speed is fast enough that it is not the cause of slowness, so any delay is all on Wolfram|Alpha."

If I didn't include that note, someone would say "Is is slow because you're on 56kbps dial-up?"

Siri and Alexa pass a lot of questions to Wolfram Alpha.

When Apple first started using it, they were responsible for 25% of all WA traffic. With Alexa, I assume that the majority of WA's queries are coming from smart assistants at this point. (https://9to5mac.com/2012/02/07/four-months-in-siri-represent..., https://www.theverge.com/2018/12/20/18150654/alexa-wolfram-a...)

I used it for Calc 1 and 2. It helped me check my work for Limits, derivatives, integrals, Reimann Summs, Series, Sequences. I love the part that says "Show Step By Step" because I can figure out which step I made an error.

The answers in the back of the book didn't tell me step-by-step how I solved the problem. It just gave me the answer and there are many times I couldn't figure out which step I made the error. Usually it was some dumb mistake, but by identifying the dumb mistake, I could remember to double check that similar step in future problems.

I had a hard time using it for Classical Physics to check my work.

Same, helped me quite bit back when I was taking Calc 1 and 2 for that same reason.

To understand the state of Wolfram Alpha, you have to understand the guy behind it.

Wolfram Alpha was a pet project of Stephen Wolfram, the creator of Mathematica. He had grand visions for it. And for the first few years, it seemed like he was doubling down on it.

But then he got bored and started tackling a bigger problem: his own solution to the "theory of everything" problem -- something that has eluded the world's best physicists for decades.

But he was confident that he could best them all. Because he created Mathematica.

The scientific community wasn't having it:

https://www.scientificamerican.com/article/physicists-critic...

I'm not sure you intend it, but your comment kind of makes Wolfram sounds like some sort of crank.

He's a leading thinker obsessively interested in this idea that everything around us is the product of a simple, fundamental ruleset.

He's sitting on the bleeding edge of human knowledge where, honestly, everyone is at risk of being full of shit. Scientific consensus isn't really any kind of indicator of future breakthroughs.

To each their own - let Wolfram be Wolfram.

For me, I never got into using it much (due to lack of experience with Mathematica syntax). I had some niche uses like "how many work days between <date1> and <date2>" but that's hardly so important.

Instead I use the SymPy Live shell https://live.sympy.org/ which does most of what I need in terms of math calculations. I'm a big fan of the sharable links (the thumbtack button below the prompt) that you can post in comments to show an entire calculation encoded in the URL querystring, e.g., https://live.sympy.org/?evaluate=factor(x**2%2B5*x%2B6)%0A%2... (factoring a polynomial), or https://news.ycombinator.com/item?id=23158095 (linear algebra helper function).

Sympy live shell is decent, and the latex rendering is pretty sweet. But, it's on ancient versions of everything, runs slowly, and has a C- UI.

Instead, I use Colab with Sympy + latex output and matplotlib (and most other things you could want to import, pre-installed). It's running new versions of things, and backed by more power, with an option to pay for even more. The latex rendering took a bit of poking around stackoverflow, but works just fine.

Feel free to copy:

https://colab.research.google.com/gist/dmlerner/23543255fdde...

WTactualF? When has * ever been anything other than multiplication? Why would the resulting links all be discussing division?

I think Google search doesn't include the special symbols, so it's like searching "48 6".

This is correct. In this query, the '*' is being disregarded. Then, I assume, more people on the internet discuss 48 and 6 in the context of long division than in the context of multiplication.

I don't know, I was looking for "how to configure cors for specific vhost in nginx" and all I got was Apache SO links. Had to use -apache.

Use verbatim search too. All words must exist without aliasing.

(google aliases ubuntu and debian, john/jon/Johnathan for example)

I use it semi-regularly; once a week or so. It's a genuinely useful tool that was just greatly oversold on launch. Things I use it for:

- Converting units while cooking. I prefer to cook by weight, and for most ingredients, you can do something like "2 cups of flour in g"

- Stuff I'd have used a scientific calculator in an earlier era: simple systems of equations, plots, etc.

- Comparing stats on countries, e.g. GDP growth in various countries

What do you mean? I used it to solve a nasty impedance network for the real and imaginary components yesterday and the solutions were accurate.

Edit: Maybe it's just good enough that people treat it as a tool and see no need to market it. It consistently has worked fine-ish for years and is useful at what it does.

My meaning was just that I saw it sometimes referenced on HN, but I haven't seen it mentioned for a while now. Hence my search and results showing 8 years since.

I guess what I should be doing is looking at the Alexa ranking of Wolfram Alpha.

You should search comments, rather than stories. It's very regularly referenced in HN comments, often for calculations, sometimes in other contexts.

https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...

I appreciate the conversation around WA this Ask HN has started, but yeah you've basically completely answered the original question by pointing this out.

They're just serving up answers which is boring to HN readers. Where's the drama in collecting data privately? Where's the drama from censoring results? No drama == No interest? Gawd, I have become cynical.

I use it exclusively when I'm drunk, to calculate how drunk I am

"4 drinks in 3 hours at 64 kg"

I think the strategy of Wolfram Research has shifted from trying to sell Wolfram Alpha as a standalone service, to selling the Wolfram Language with WA functions for retrieving standard datasets. A finance professional, for example, probably did not gain much information from asking WA "would it be better to invest $100 in GOOG or FB in 2013?", but the `FinancialData` function for pulling end-of-day stock prices enabled these people to do interesting analysis that they couldn't have done otherwise.

(source: conjecture, but I did work at WR for 3 years and on the initial Wolfram|Alpha release)

My guess is that it’s a bit too complicated/slow for a lot of ordinary people and too finicky for a lot of technical people.

I’m a frequent Mathematica user and I find almost all of my use cases require several different attempts to get the desired result w/wolfram alpha. Meanwhile, most people who don’t get the right result the first time will probably just give up and not think to rephrase the query.

It just can't answer the questions I have. Last time I tried it, I was looking for buoyancy of various gases, but it insisted any such question necessarily referred to stuff on water.

It did OK figuring the fake "temperature" of LHC beams that fusion people like to quote because they sound more impressive than GeV.

I still use it regularly too - even more so after listening to Stephen Wolfram’s 3 part podcast [1] with Lex Fridman where he discussed the latest developments in Wolfram, Mathematica etc

[1] https://youtu.be/ez773teNFYA

It's still around but I imagine it is experiencing a bunch of competitors biting chunks out of it.

A lot more people can script now, so open source packages of computer algebra systems (Sage, numpy, scipy etc.) Probably take a small bite.

And then you have closed source ones to consider like Matlab.

The second largest chunk probably being bitten out of it is its web and app competitors (desmos, symbolab, etc.) Alexa rankings show that these see a lot more traffic and engagement (2 - 3 times).

Finally, a small portion of its functionality is now covered by search engines. I imagine they'll continue to gobble things up. There are also a few good Web tools, I used one for a linear algebra course I found a lot better than the freeware version of WolframAlpha that came with my Raspberry Pi.

I can't find any reports on its revenue or net income. I would be super curious who uses it. Maybe it's growing... who knows? I also remember it being recommended a lot in the early 2010s.

You are mixing things up here. The headline is about Wolfram Alpha. You are talking about Mathematica.

I'm talking about both. When I was comparing them to competitors like Symbolab I was using the Alexa ranking for alpha.

I find it faster and more accurate to use a specific package in an interpreter than query Wolfram Alpha or use Mathematica. And for the simpler things a search engine will do!

I mainly use it as an english dictionary of math terminology.

Although for the basics of differential geometry like the Weingarten equations and the Dupin indicatrix WA is lacking - as is Wikipedia except for the articles in the german Wikipedia. And I haven't found a way to get to the 'Weingarten equations' searching for 'Weingarten', you only find him by the full name 'Julius Weingarten'. :(

https://de.wikipedia.org/wiki/Weingartenabbildung https://www.wolframalpha.com/input/?i=weingarten+equations https://de.wikipedia.org/wiki/Indikatrix https://www.wolframalpha.com/input/?i=dupin+indikatrix

If only there was some way to contribute to the english wiki (article or search) when you are lucky enough to understand the (better) german one...

As someone who just signed up for an open university this semester, I'd love to hear opinions about Octave and Maxima for general purpose use. Especially for study, such as replacing Wolfram Alpha's step-by-step solutions.

I'm a Linux user and prefer an open-source solution. But I have no objection to paying a reasonable amount of money for a good commercial solution. Maybe Maple is worth looking at?

Honestly, Google can now do most of the basic things that WA could do.

And the more complex things WA could do oftentimes require a bunch of trial and error to figure out the correct syntax/phrasing to use to get correct results, to the point where it was just easier to either do the calculation manually or find a dedicated site for it.

So it has just lost utility for me.

I was enthusiastic, but for medium complexity questions I spend more time footing with syntax then it would take to do it myself. I probably use it for a high complexity question once every few months. I’m happy that it exists, on balance

Nothing. It's still solving my homework. (Sometimes.)

I use it regularly, like twice a week.

When I'm making exercises to explain to my students in the math class, I use W.A. to double check the answer.

I also use it for calculation for comments in HN. Sometimes I need to make a back of the envelope calculation, and W.A. can convert the units and other boring stuff.

It's fun for life expectancy.

Step one: Ask for your own life expectancy.

Step two: Ask for the life expectancy of someone years' younger.

Step three: What.

Step four: Oh.

Huh? This was not at all surprising, someone younger than me had a lower life expectancy, while someone older was higher

They put their "step-by-step" explanations behind a [login/pay]wall which made it significantly less useful.

Out of sight, out of mind. It's still there

I think they've always been like that.

Good thing is, they have a montly cost, but the mobile app you just buy once and it works forever. And it's not that expensive iirc.

This sounds misleading, the app offers Pro subscription for $7/month.

Step by step solutions were free around 9/10 years ago

  > They put their "step-by-step" explanations behind a [login/pay]wall which made it significantly less useful.

Maybe, but what else can do step-by-step explanations? Perhaps octave?

I use it whenever I have something mildly annoying to convert, especially dates. e.g. https://www.wolframalpha.com/input/?i=1636221900+unix+time+i...

Probably an incredibly trivial use-case but still useful regularly for me...

It doesn't work so good for times but I often use Google search to multiply numbers with units together and get a result in the units I want without having to worry about screwing up unit conversions.

Example: 4 atomic mass units * (1000 nm/sec)^2

Google Result: 6.64215616 × 10-39 joules

I use this all the time. I use wolfram alpha for solving equations or systems of equations but I use google for unit conversions because it's got better input parsing (frankly).

I should try the wolfram alpha math entry mode probably, I think that didn't exist when I started using it. If I could manually enter the equations with stricter formatting to ensure it's interpreted properly I'd use it more.

A reminder that GNU Units still exists, e.g.

  $ units
  Currency exchange rates from FloatRates (USD base) on 2021-01-17 
  3677 units, 109 prefixes, 114 nonlinear units
  
  You have: 4 amu * (1000 nm/s)^2
  You want: joules
          * 6.6421563e-39
          / 1.5055352e+38

  You have: ^D

It’s slightly less DWIMish (you have to say “atomicmassunits”, “atomicmassunit”, “amu”, or “u”, not “atomic mass units”) and somewhat awkward as a separate tool, but then resorting to your web browser for unit conversions is awkward in a different way. Non-interactive invocations, like units VALUE-OR-UNIT UNIT, work as well.

[1]: https://www.gnu.org/software/units/

+1 for 'units'. I like it for conversion between millilightseconds and miles, to get the theoretical best-case latency between two places.

i.e. if it's x milliseconds ping, it can't be more than m miles away.

I can’t tell without checking carefully if you’re serious or trolling (cf. the Borwein trig integral troll), but that’s the point I suppose. And I don’t have any visceral intuition for miles anyway, apart from “kind of a bigger kilometre”, even if I remember the conversion factor (having been taught SI in high school and Gaussian CGS in college).

A similar approximation that I’m sure works :) is speed of sound in air ≈ 1/3 km/s, so you can count the seconds from a lightning strike until you hear the sound and divide by three to get distance in kilometres.

I bought Mathematica and I'm using its free text input instead. But I guess behind the scenes, it's WA again.

I only ever use it for date math

For whatever reason, I like keeping track of 1000 day anniversaries

https://www.wolframalpha.com/input/?i=1000+days+after+today

Shortly before any kind of 3rd anniversary or birthday I try to remember to check this.

You might like this little toy website I made a couple of years ago:

https://interesting-anniversaries.com/

From my readme:

“Have you ever wanted to know when you turn 2 billion seconds old? How about 33,333,333 minutes old? When do you get to celebrate your 555,555th hour of life? As it turns out, all three of those milestones occur in the same 24-hour period!”

I use it to solve differential equations.

The name makes it seem like pre-beta test software.

I'm waiting for the final release, and then I'm waiting some more for it to be declared stable, and then I'm waiting some more for it to catch on and be declared popular.

Not really, but that's what the name suggests to me.

I just tried it here because of TFA and it's good.

I think students these days use it for math/calculus, but it isn't seen as something special because they've always had it. It wasn't novel like for us.

I used to use it a lot but google now provides most answers as well and much faster. Wolframalpha performance is still sluggish and 6 second loading for a bunch of text (simple queries like `6cet to pst` is frustrating)

At University pretty much everyone I know uses it for homework.

Since nobody is mentioning it, around that time Wolfram Alpha started paywalling a lot of the more useful features. I used to use it in school and stopped when that happened. I'm not sure if they changed course since then.

I still use it once in a while when I don't want to bother converting non-base10 units, like to know the date in 90 days, or how many hours in x days, etc.

Used it a bit at university to compute some complex integrals if I was stuck or feeling lazy. That was 11 years ago.

Don't think I've even visited the website in the past 6 years.

It did about a lot of the heavy lifting in my master's thesis, not gonna lie.

It’s still there, and I use it regularly, probably several times per week.

Used a lot by math students to check answers.

I still use it frequently for any random calculations.

I use WA for complex math at least once a week.

> I did a search on comments on HN for Wolfram Alpha. Most posts are 8 years old, none newer, some older.

You searched wrong. Excluding today, the most recent comment was 7 days ago, and there were quite a few more in the past month.

https://hn.algolia.com/?dateEnd=1636070400&dateRange=custom&...

I still use it all the time fore unit conversions, odd time based questions, etc. I find it's way better than the Google results because if I think of something after the fact I can tack it on and WA figures it out better than Google. E.g. "12 ft to meters * 3" is not handled right by Google but is handled how I want by WA.