Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Wolfram Alpha doesn't "get it"

blog.agiletortoise.com

Post reply on HN
Print view View original

81–90 of 111 posts

Re: Wolfram Alpha doesn't "get it"

#81

Post by malnourish »

steveklabnik wrote: post #68
bpodgursky wrote: post #50

I think it's important to note that many queries on Wolfram Alpha are VERY heavyweight--I've had complex calculations and rendering take 30 seconds to a minute to finish. It's not like Google where a query get a result in a few milliseconds. So while Terminology is not likely to be querying it with lots of formulas, I think the policy exists so a badly thought out automated querier can't accidentally put a huge load…

There's an easy technical solution to this: just block the freaking querier. Also, if they didn't render a result as an image, maybe it wouldn't take so freaking long...

Now, I am probably wrong but am asking this due to curiosity. Could they conceivably use HTML5 to render dynamically thus reducing resource/load cost?

Edit: No to Now

Re: Wolfram Alpha doesn't "get it"

#82

Post by steveklabnik »

malnourish wrote: post #81
steveklabnik wrote: post #68

Earlier quoted context omitted.

There's an easy technical solution to this: just block the freaking querier. Also, if they didn't render a result as an image, maybe it wouldn't take so freaking long...

Now, I am probably wrong but am asking this due to curiosity. Could they conceivably use HTML5 to render dynamically thus reducing resource/load cost? Edit : No to Now

I hope they don't try to sue me for this post.

    http://www.wolframalpha.com/input/?i=2+%2B+5
(oh noes!) gives an image containing '7'. They could just freaking put a 7.

Granted, not all of their queries could be done this way, and some of those could indeed be done with canvas, for example. Dunno if those would be faster. But all of the numbers and tables could stand to just be numbers and tables.

This is more of a gripe about copy/paste than performance, really: I'd imagine their calculations take a lot longer than making a .png

Re: Wolfram Alpha doesn't "get it"

#83

Post by ars »

malnourish wrote: post #81
steveklabnik wrote: post #68

Earlier quoted context omitted.

There's an easy technical solution to this: just block the freaking querier. Also, if they didn't render a result as an image, maybe it wouldn't take so freaking long...

Now, I am probably wrong but am asking this due to curiosity. Could they conceivably use HTML5 to render dynamically thus reducing resource/load cost? Edit : No to Now

The rendering is not the hard part - the mathematical calculation is.

Re: Wolfram Alpha doesn't "get it"

#84

Post by foljs »

perlgeek wrote: post #64
tptacek wrote: post #60

Earlier quoted context omitted.

Tweeting a link seems like a good example of a "common sense use of the website" that could bring it to its knees if some queries take 30+ seconds to complete.

Caching exists.

How about a query that calls a function involving random numbers?

Re: Wolfram Alpha doesn't "get it"

#85

Post by foljs »

steveklabnik wrote: post #68
bpodgursky wrote: post #50

I think it's important to note that many queries on Wolfram Alpha are VERY heavyweight--I've had complex calculations and rendering take 30 seconds to a minute to finish. It's not like Google where a query get a result in a few milliseconds. So while Terminology is not likely to be querying it with lots of formulas, I think the policy exists so a badly thought out automated querier can't accidentally put a huge load…

There's an easy technical solution to this: just block the freaking querier. Also, if they didn't render a result as an image, maybe it wouldn't take so freaking long...

> Also, if they didn't render a result as an image, maybe it wouldn't take so freaking long...

Parent got upvotes? Isn't HN is supposedly populated by technical people?

Rendering the result as an image is trivial and fast in a modern server.

I very seriously doubt this is the problem, instead of doing all the symbolic mathematical calculations in the first place...

And, of course, the reason to render the result as an image is to have a consistent cross-(legacy)browser support for mathematical notation. Wikipedia renders mathematical formulas as embedded images too. No other solution (from MathML to Flash/Canvas etc renderers) is as widely supported (and none is as complete as Mathematica's math rendering engine).

Re: Wolfram Alpha doesn't "get it"

#86

Post by malnourish »

steveklabnik wrote: post #82
malnourish wrote: post #81

Earlier quoted context omitted.

Now, I am probably wrong but am asking this due to curiosity. Could they conceivably use HTML5 to render dynamically thus reducing resource/load cost? Edit : No to Now

I hope they don't try to sue me for this post. http://www.wolframalpha.com/input/?i=2+%2B+5 (oh noes!) gives an image containing '7'. They could just freaking put a 7. Granted, not all of their queries could be done this way, and some of those could indeed be done with canvas, for example. Dunno if those would be faster. But all of the numbers and tables could stand to just be numbers and tables. This is more of a gr…

Makes sense.

What if they output in LaTeX for more complex things, and there was a LaTeX plugin for browsers?

Re: Wolfram Alpha doesn't "get it"

#87

Post by runningdogx »

jwr wrote: post #25

Seems to me the author doesn't get it. Wolfram Alpha specifies terms of service, which the author didn't read and subsequently violated.

You read every ToS for every service/product you use?

Community and social standards often trump legalese. For instance, most absurd restrictions in apartment rental agreements are not enforceable.

In this case, linking to another site, no matter how "deep" the link is, is common practice.

If WA really doesn't like it, they can prevent it by checking the referrer, or going even further by adding a pseudorandom token to their query page, switched every 5 minutes and valid for an hour or a day, and redirecting deep links which don't have a valid token.

Of course, doing that would probably kill some traffic they like, so they have a silly, selectively-enforced and probably unenforceable ToS clause instead.

I don't see any reason why such provisions need to be enforced legally, since they can be enforced technologically in most cases. Expanding the law because of laziness is a horrible precedent.

Re: Wolfram Alpha doesn't "get it"

#88

Post by wnoise »

jwr wrote: post #25

Seems to me the author doesn't get it. Wolfram Alpha specifies terms of service, which the author didn't read and subsequently violated.

It's not clear to me why the author should be bound by terms of service that he never agreed to.

A web request is merely a communication with the web server. The server is free to do what it wills with such a request. If Wolfram doesn't want to respond to requests from the author's apps, that's well within their rights. Similarly if they do respond, it's because they explicitly set it up to so respond.

Re: Wolfram Alpha doesn't "get it"

#89

Post by 27182818284 »

malnourish wrote: post #86
steveklabnik wrote: post #82

Earlier quoted context omitted.

I hope they don't try to sue me for this post. http://www.wolframalpha.com/input/?i=2+%2B+5 (oh noes!) gives an image containing '7'. They could just freaking put a 7. Granted, not all of their queries could be done this way, and some of those could indeed be done with canvas, for example. Dunno if those would be faster. But all of the numbers and tables could stand to just be numbers and tables. This is more of a gr…

Makes sense. What if they output in LaTeX for more complex things, and there was a LaTeX plugin for browsers?

I've seen Chrome plugins and I know there are Greasemonkey scripts for automatically LaTeXing plain text on web pages.

I don't think this is a viable solution for W|A, however.

Re: Wolfram Alpha doesn't "get it"

#90

Post by blahedo »

dpcan wrote: post #28

You break the rules, you pay the consequences. Ignorance is NOT an excuse. He's "in his situation" (edit) with Wolfram because he didn't do his due-diligence and contact them first before linking to them in his app. EDIT: He seems to understand this, and admits that he "goofed", yes.

> You break the rules, you pay the consequences. Ignorance is NOT an excuse.

While this is technically true, I feel that it has slipped to the level of a polite fiction rather than a core civic principle. People pretty much never read contracts anymore because they've gotten so long and dense, and the contract writers know this and make them longer and denser anyway. Apple, among others, presumes you won't actually read their agreement before clicking through---they don't bother to sync up the printable version, don't provide a diff, and their system times out before you could possibly have read even a small part of the document. (I've written about this before, at some length: http://www.blahedo.org/blog/archives/001060.html .)

Post reply on HN