Hacker News Reader

Comment by barik | Hacker News Reader

This article was light on details, but I thought it was important to address at least the first point, which has a fundamental misunderstanding right off the bat:

"1. Plagiarism is bad.

Wrong! If the licensing is right, copy to your heart's content. If you're not in violation of copyright, trademark, or patent, you can do whatever you want with someone else's creation. Sometimes they even give you permission. Did you find a real swell formula online? Did you know you can't copyright formulas?""

This isn't plagiarism. Plagiarism is when you copy others' work and claim that the work is yours. You are welcome to copy the entirety of a Wikipedia article verbatim and submit it, as long as you appropriately cite it. Of course, you may not get a good score, since your contribution to the work was zero, but you haven't done any unethical.

This is true both in academia and industry.

View on HN

Most open source licensing requires 0 citation. If I copy your formula or recipe, I don't need to cite a thing - you don't own the copyright to your own formula. I could sell your work if I wanted to do so.

Plagiarism, as it affected my grading, was copying without attribution. Attribution is rarely required for most licensing, except where it is. These sorts of distinctions are lost in the collegiate machine.

Which open source licenses are you talking about? The most common that I can think of is GPL, which does not allow you to remove the copyright notice, although you may add yours in addition to it. The next one that comes to mind is the BSD license, which states:

"Redistribution and use in source and binary forms are permitted provided that the above copyright notice and this paragraph are duplicated in all such forms and that any documentation, advertising materials, and other materials related to such distribution and use acknowledge that the software was developed by the <organization>."

Whew. So maybe the third popular might be The Apache License, but this states:

"You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to..."

And so on. I am unaware of, but would be interested in knowing about, licenses that would allow you to verbatim take code and also claim it as your own work.

I'm almost always using binaries or javascript libraries, in which I need not do anything but copy and paste files and include them in my projects to be within the letter of those licenses.

Plenty of people copy code straight off of blogs, which due to the DMCA are given implied copyright. Stackoverflow, also, requires attribution under the CC license, but nobody attributes stackoverflow when applying fixes to their problems.

If I include an MIT licensed JS file, I can replace functions willy nilly, or rewrite parts of the code. If it's LGPL3 or GPL, I can't.

Licensing is more nuanced than you're describing, but pragmatically I never need to cite anything. The most I need to deal with licensing is handling proprietary code (unlicensed github code, for example), and GPL/LGPL.

When you "include an MIT licensed JS file", as you mention doing, you leave its copyright/license notice intact in the source right? That's the acknowledgement required. You didn't change the copyright to have your own name on it and erase the original.

Here is the WTFPL, which is my favorite license as it works as well as public domain, but, being a license, deals with the concerns of people who believe public domain is a legal fiction in certain jurisdictions.

http://en.wikipedia.org/wiki/WTFPL

This is the only total freedom license I am aware of, and allows claiming others work as your own if you wish.

I agree the OP was confused about the word plagiarism, but from his follow up it seems clear he meant not giving acknowledgement, as opposed to falsely taking credit. Regarding non-acknowledgement and non-advertising clause licenses, the zlib is perhaps the most popular and is a really nice license. It has an anti-plagiarism clause that you can't falsely claim it is yours, but you are not required to acknowledge ownership in binary distributions. (It is appreciated but not required.)

http://en.wikipedia.org/wiki/Zlib_license

There is also the anti-copyright notice, which varies in form, but is an explicit public domain grant required by the Berne Convention. Typically they do not require acknowledgment either.

http://en.wikipedia.org/wiki/Anti-copyright_notice

What I wonder is how far copyright protection goes, specifically when AI algorithms are involved. Suppose you "copy" a recipe + description (or a book for that matter) in the following way :

1) you create markov chains using a large number of cookbook texts

2) you create the markov chain for a specific recipe's description and go down alternate high-probability paths in an effort to creatively change the wording without changing the content (maybe this is comparable to the way a human would if he tried the recipe and then wrote down "how he did it")

3) you republish the result

Did you violate copyright or not ?

The reason I ask is that this is often used as an end-run around patents and copyrights. What most speech recognition programs do these days is preprocess the sound + feed into neural network + get output. (and while most "pattern recognition" algorithms for non-visual things seem to have a love affair with support-vector machines, temporal neural networks are certainly advancing there too).

Now if you analyse what those neural networks do there's 2 types of things 1) ~40% effectively is unrolled loops of (mostly) patented algorithms 2) 60% you effectively don't recognize (to be fair it takes hours of seeing the network operate before you realize anything it's doing)

Clearly this is legal, in cases human implementations of the exact same algorithms wouldn't be, nicely sidestepping the problem of patented algorithms. And as a bonus you don't really have to know the subject matter (e.g. you can write a pretty good voice recognizer without 8 years experience as a linguist. Or you can write them for languages you don't actually know). And as a bonus, academics are miles ahead of the private sector where it comes to machine learning algorithms, so extremely useful thing are effectively free-for-all.

Of course this is also what humans do. If you look at neural networks, they can only do what they've "seen" happen before, or they can combine various things they've seen before. But they are utterly incapable of coming up with original work. So I don't think humans are any different to machines when it comes to producing original work based on combinations of previous works. This doesn't mean the thing that was copied was itself copyrighted, you can write about your own life, for example, or about nature, or ... but we'd call that original, when (in a strict mathematical sense) it's not.

So clearly we've de-facto accepted in our society that machine-processed works at some point start constituting original work.

Do we have any data what point that is ?

Any Ethics course as required for ABET accreditation would have taught you that copyright is substantial in software. Most open source projects require the retention of the copyright notice at the top of every file.

As far as I know the two most popular licenses are MIT:

"The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software"

And GPL:

You may convey verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice;

The BSD license would like a word with you:

"Redistribution of source code must retain the above copyright notice ..."

Or how about the Apache license:

"You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work,"

Maybe the MIT license:

"The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software."

and so forth.

Coming from Lua, where everything is MIT:

"substantial license"

If you copied a couple of functions for a class the uni would have big problems if you provided no attribution; though the license does not require it.

I think you may be confused as to how OSS licenses work, and to a larger extend the OSS community. I'd recommend you read "The Cathedral and the Bazaar". On the specifics of license, I guess good ol' Wikipedia works just fine: http://en.wikipedia.org/wiki/Open_source_license.

Now your teachers didn't focus on entrepreneurship because statistically speaking, most people who go through college (or any kind of education) will work for somebody else.

I'd add to this that school is, in most cases, not about learning practical skills. It is about learning how to learn, developing mental tools and acquiring a knowledge foundation that will help you work with others.

Debatable approach to education, maybe. But beware of thinking that everybody want or should be prepared for the career path you seem to be in favor of.

"As it affected my grading"...

Reply on news.ycombinator.com