[phkahler]

Google is failing to organize the world's information in a way that is useful to me. I wanted to look up how to split a NURBS surface in half. You know, the math behind computing the new point and weights. All I got was a bunch of stuff from CAD vendors talking about how to manipulate NURBS in their own tools.

Since then, I've seen more of this trend. It's hard to find information about how anything works now because most the results are commercial - SEOed to be high in the rankings.

I'd like an option to screen out commercial results in favor of more informational ones. They're supposed to be helping me find information right? But in reality they've shifted to feeding me marketing information.

[VikingCoder]

Splitting a NURBS surface "in half" is an insanely complicated operation, and the way you do it almost certainly depends on your intended use case.

If you can't find a library (or engine) in your domain that has the operation, then you're kind of S.O.L., and have to start from first principles.

I don't blame commercial results and SEO for this. It's not like you used to be able to see a menu of rich choices for this, and now they're lost in the noise of companies with a mission.

[jlarocco]

> Splitting a NURBS surface "in half" is an insanely complicated operation, and the way you do it almost certainly depends on your intended use case.

It's actually not "insanely complicated", and there are resources around, like https://pages.mtu.edu/~shene/COURSES/cs3621/NOTES/spline/NUR... and https://www.amazon.com/NURBS-Book-Monographs-Visual-Communic...

> It's not like you used to be able to see a menu of rich choices for this, and now they're lost in the noise of companies with a mission.

I mentioned just yesterday that DDG results are better for me, and this case is an example. "NURBs splitting algorithm" turns up a bunch of results.

[VikingCoder]

> It's actually not "insanely complicated"

Hi, your first link talks about NURBS curves, not NURBS surfaces. Cutting a surface in half is significantly more complicated than cutting a curve in half.

And your second link is to a commercial result, an entire book. Presumably because the topic is complicated enough that people are willing to pay $69 to $78 to understand the challenge well enough.

If your DDG search for "splitting a NURBS surface in half" are successful, I'd love to see the results.

[phkahler]

>> Cutting a surface in half is significantly more complicated than cutting a curve in half.

No, it's actually the Exact same thing. Just applied independently to each of the curves in one direction.

But none of that is really important. 15 years ago it would be trivial to learn how to do it with a Google search. Today not so.

Also, a book result is still better for me than the commercial tool links which contain no math or theory what so ever.

[VikingCoder]

> No, it's actually the Exact same thing. Just applied independently to each of the curves in one direction.

I can easily imagine situations where I would not be happy with that result.

If that works for you, that's great.

> Also, a book result is still better for me than the commercial tool links which contain no math or theory what so ever.

Searching Google for "NURBS Book", "NURBS Course", "NURBS filetype:pdf" all work great for me.

At a guess, most of the people who use Google are using tools, not making them. Yes, Google favors results that make the majority happy. I see that as a feature. If I'm in the minority, yes, I do need to become a more sophisticated searcher.

[lostmsu]

I am using DuckDuckGo too, but to be fair the first result in Google on that exact query seems relevant: https://computergraphics.stackexchange.com/questions/340/spl...

[m463]

A better example would probably be to search for a way to fix something on your car. The results are overwhelmingly skewed towards selling you the part vs instructions.

[jacquesm]

Yesterday I wrote a blog post about an interesting company, I couldn't find the URL even though I knew exactly what they did and where they were located. It took forver - and DuckDuckGo - to finally find the website. I'd forgotten their name and had put of writing that blog post for long enough that it had slipped my mind.

That isn't the first time this happens either. It's extremely frustrating to not be able to locate a page that you know is out there.

[shantly]

I used to be able to find specific pages on the Web by googling for unusual words I expected would be used to link to them, not on the page itself, to pick an extreme example of the sort of triangulation one used to be able to use to track down lost pages or sites.

That kind of thing hasn't worked in... god, at least a decade, at this point. Not just because the Web is larger, but because search doesn't seem to work like that anymore and also everything got way spammier. "Clever" searching is a skill I was once (judging by people's reactions) notably good at but that is entirely obsolete, but not because the function it served was replaced by something better—it's just gone now.

[gowld]

Was this not a front-page result, as it was for me?

https://computergraphics.stackexchange.com/questions/340/spl...

[phkahler]

That turned up at some point. But note that 1 it doesnt give an actual answer. And 2 I didnt need to read about knot vectors at all, so it wasnt the correct solution for me.

[magashna]

May be better to select by org type (site:.edu) or remove what might be obvious commercial results (-buy -"free download")