26 comments

[ 2.3 ms ] story [ 71.1 ms ] thread
This kind of shit is why we can't have nice things. Software patents' only use is as cudgels for large companies to beat other large companies with, and they leave a hazy minefield in their wake for software developers to wade through.

Abolish software patents.

https://endsoftwarepatents.org/

(comment deleted)
The whole point of trademark and patents are to encourage innovation and creative production, otherwise, it would be a waste of collective resources or unfair restrictions at the very least, if they didn't result in more investment in research and development.

You're being down voted, but I am curious if there is any software product or enterprise in the market that couldn't have been competitive (not a monopoly!) without patents?

maybe mp3?
Patents on codecs are probably the worst offenders in my mind. Codecs benefit significantly from network effects, so making it a question as to whether you can legally use a codec muddies the waters and drives people away from good solutions. You wind up in ridiculous situations where you have hardware in your CPU that you can't legally use without paying for a license.

Plus, there are so many parties with a vested interest in making good codecs, I really don't buy the argument that they require patent protection otherwise they wouldn't be researched.

With the current level of investment in video codecs, it takes about 6-8 years to achieve a 40-50% BD-rate improvement between codec generations. That amounts to thousands of inventions over the course of those years. Without patents, I bet the level of progress would be abysmal.

What vested interest are you referring to? What incentive is there for companies like InterDigital, Ericsson, Nokia, or even Qualcomm to spend hundreds of millions of dollars on each generation of codecs?

It takes a lot of money to sustain research and development in video codecs. A decent researcher makes between 100-300k USD a year. Each generation of H.26x requires hundreds of researchers. JVET meetings have an attendance of about 350 people, four times a year.

And don’t get me started on the computational side. A single research team needs thousands of CPUs running 24/7 to sustain the research. Video codecs are seriously slow during development. They get fast once dedicated hardware is built.

As a video coding scientist, it baffles me how people assume that our work should be free. We work enormously hard for years so that YouTube and Netflix can stream higher quality video for everyone to watch. Now, why should that be free? Why do you call our work shit?

> What incentive is there for companies like InterDigital, Ericsson, Nokia, or even Qualcomm to spend hundreds of millions of dollars on each generation of codecs?

If there's no incentive then they are the wrong companies to do the work.

Google, Netflix, Amazon, Apple, Microsoft, Meta, etc. all derive clear and direct benefits from video codec improvements. They have strong incentives to reduce the bandwidth usage and increase the image quality of internet video. That's why they've all contributed to and use AV1:

https://aomedia.org/membership/members/

All protocols and formats on the web are available to implement under royalty-free terms:

https://www.w3.org/policies/patent-policy/

Audio and video are no different. If H.265 can't meet that standard then it's the wrong choice for the web.

AV1 for video and Opus for audio are formats that achieve royalty-free licensing and are better fits for the web and the internet generally.

How far would we be if darpa net/ip networking was under a patent, the DNS system under a patent etc. etc.
What exactly was Netflix doing that infringed? Using AV1? H265?
It's in the text: HEVC/H.265
Do the shareholders get to vote for imprisonment instead? As surely that is the more preferable option for most of them...
I don't think Netflix's board going to prison would benefit their shareholders very much
what?
>The court set the fine at €150,000, or in the alternative 15 days of imprisonment for members of Netflix’s board of directors, for each of the 47 days that Netflix had infringed in violation of the cease-and-desist order.

Implying that imprisonment for board is a real option.

Why Germany? Both Netflix and Broadcom are American companies, so isn’t it more logical to pick a venue in the US? Was Broadcom’s case so flimsy (with regards to US patent law) that they had to find an overseas jurisdiction to get a favorable ruling?
The case in question is about a patent registered in the European Patent Office, which is headquartered in Munich
(comment deleted)
From the linked post:

“PALO ALTO, Calif., Dec. 22, 2023”

So quite old news, IMHO? Shouldn’t that be mentioned in the subject?

I implemented a morton/z-order curve, and implemented the LITMAX/BIGMIN:

https://en.wikipedia.org/wiki/Z-order_curve#Use_with_one-dim...

It helps implement spatial partitioning with a z-order curve, which makes spatial partitioning quite fast and easy to implement compared to R-tree, quad tree, KD tree and other sophisticated things.

The wikipedia article doesn't mention if it's under patent, and some link says the patent was dropped, but I am not sure.

https://patents.google.com/patent/DE10331679A1/en

I don't really know how to read this, and there are other results to "LITMAX/BIGMIN patent".

Although, to be quite honest, I don't think there is a good benchmark of spatial partitioning algorithms (like there is for hash tables), since it's often comparing apples/orange/bananas.

(comment deleted)
(2023)

One of a number of Broadcom releases about the case. Has there been any change/appeal from Netflix since?

Some more discussion around then: https://news.ycombinator.com/item?id=38249527

I am sure they saved enough in 40 days to pay for the fine.

Also, while I have no love for these patent pools, the technology is clearly commercially restricted to license holders. Netflix should either license or stop using it.

Switch to VP9 and AV1?

And this is why Netflix is in such a bitter fight with Roku over AV1 support.