Genius Scheme Catches Google in the Act

What an intro! 😅 thx, will be fun. ✌ grt team

I thought it would be something to do with non-breaking zero width spaces or w/e but this is way better.

Interesting. The app closed unexpectedly while watching this video….sketchy.

It's More Skode!

It's a shame this will never work again; if Google learned anything here, is to normalize all characters to a specific set of characters whenever they steal more content.

I think Google will just turn this over to the music labels and let them go after ALL of the content on their website and let the music labels just crush Genius as all their content belongs to music labels. Genius has more to lose than Google.

A comment in the chat replay says it is not possible to avoid using Unicode because of non-ASCII characters that naturally occur in lyrics. That may be true, for lyrics not written in English, such as in Spanish, French, German, Scandinavian languages, etc.

I referred in another post to replacing exotic characters by standard ASCII ones. ASCII is Unicode codes U+0020 thru U+007E. However the usual coding for Western HTML web pages is the default one, which uses U+0020 thru U+00FF for printable characters. Though it does not in theory use U+0080 thru U+009F, in practice it does, as Microsoft very conveniently provided printable characters in that range, and those are now in the HTML standard.

Along with the U+00A0 thru U+00FF characters, they cover everything needed for a wide range of languages (Western Europe, North and South America, and the Anglosphere countries), because they provide good support for accented characters.

Converted to use the U+0020 thru U+00FF range, there is little opportunity for watermarking. One possibility is the use of hard spaces (U+00A0) in place of normal spaces (U+0020). This is potentially useful because lyrics contain plenty of spaces. And who is going to notice extra ones at ends of lines, for example? But of course stripping trailing spaces and tabs, and converting hard spaces to normal ones, is trivial for a programmer.

As a software developer I can tell you that it is trivial to write some code that will convert exotic unicode quote characters to standard neutral single and double quote characters available in standard US ASCII. Similarly with other exotic punctuation characters.

That would get rid of such a watermark easily, leaving only other tricks, such as use of redundant space and tab characters, say at trailing ends of lines. Just saying. This is not an invitation to do so, just a statement of the totally obvious (to anyone who can program, as they will know plenty about Unicode and ASCII).

Another approach would be to display the lyrics as a graphic. It is a lot easier to hide watermarks into images. Of course then copy-paste does not work unless Google or others were to employ OCR to decode the text from images, which is actually a known practice in classifying images by any textual content they contain, very useful if you want to be able to search for them by their incidental text content.

Google's steamroller can ignore copyright, because it would be a brave copyright holder who tries to stop their work coming up in search results, because Google can make sure they lose out by not getting the publicity that comes from being in the search results, publicity that might drive people to the copyright holder's site. Of course if Google is NOT letting people get to the copyright holder's web site, that is a different matter, but then again, who is going to take on Google?

Using Google to drive customers to your own website, hmmm. Isn't that what Google is for? To get on Google search listings so people can find your website. Banning you for doing that appears to be an anti-trust violation.

Man, the law is so unfair sometimes. This entire case was dismissed for failing to state a claim. Because the judge interpreted all of these violations as copyright.

But covering up your shady practice is more than a copyright violation.

I love watching this guy while I sit here and snort my counterfeit pain pills (Fetty) because medical industry is so against pain patients that we have to find our drugs on the street. FYI my left hand was destroyed in a car accident and had to be reconstructed with titanium joints and rods and I have to crush vertebrates in my neck and they still will not even give me a Vicodin.

I love that you gave a shout out to AvE. I'm wearing his tits and pickles shirt right now and since I support him, I'll support you.

Great video. A+

I thought that lyric box was sus when it first started appearing... but what's even better than being validated by this was the sheer creativity of that watermark! Embedding it and scraping Google search results for it? Cheap little toy script, could probably write it and make it run in about a day, between other work. Using special Unicode to mark the thing, in Morse code no less? That, my friend, was glorious! I wish I could put something like that in my performance reviews.

Hey Leonard, what's with the funky porn groove?? Gaaaroooooooooovy man! You guy's swing??? (<:

Prime example of "never interrupt your enemy while he is making a mistake". At the beginning I was thinking this was pretty stupid but seeing how Google tried to hide their shenaniganery led me to lose all sympathy.

Ok. Here is what I don't understand. Yes, the copyright to the lyrics to the song itself are owned by the person who wrote the lyrics or the company that employed the person who wrote the lyrics. That is not up for debate. However, the way that I see it, this debate is not about the song or the lyrics, but about the time and effort involved in transcribing those lyrics. As far as I am aware, the lyrics for a song are not automatically available when a new song is created and released to the general public. This means that third parties have to expend their labor to analyze the song and transcribe the lyrics. At this point in time, and I would argue at no point in time (even in the future), will there exist software that will be able to automatically transcribe song lyrics provided an audio sample. It is just not possible. Such technologies may be able to provide an initial outline, but manual analysis of the song and corrections to the transcription will be required. That being said, is one not entitled to the fruits of their labor? The whole reason that this was an issue in the first place was because Google was unwilling to do the work themselves, meaning that their is cost associated with that labor. Put another way, that labor has value. The way that I see it, Google stole from Genius in the form of the labor that went into making the service available, as outlined in their terms of service. The song lyrics and song itself have little to nothing to do with this in my opinion. This is about theft of labor, and Google should absolutely be held accountable for it. But I am not a lawyer. Would love to know what you all think.

Oh my God. THat dog is GORGEOUS. Completely stole the show. Sorry French.

Its not human contract law, its human-centipede contract law. Leonard, you need to examine your internal human-centipede biases.

Looking up how this case went, pisses me off on several levels. As far as I can tell, the federal court basically ruled that facts which implicate federal law cannot cannot be tried under state law. So, in this case, because the copying implicates copyright law, there can be no damages from the copying other than copyright damages, which Genius did not have standing to claim.

Maybe someone can correct me, but this logic would seem to say that a state or local authority can't enforce zoning regulations on land if the EPA can require an environmental impact review.

This is very similar to a tactic used by map makers to catch thieves. They'll include fake streets or even towns and when another map maker copies it it doesn't matter that the information on the map is public domain, because the fake stuff is a creative work.

Pure genius! They are brilliant

Update: It was all thrown out for failure to state a claim.

I really hope that whoever came up with that sneaky watermark method got a nice raise and bonus for that incredibly clever idea.

Genius IN DEED!

And still, Google shows no remorris for their actions.
(The misspelling is intentional.)

Its a trap street making it original copyright material. In my opinion.

one could argue that they have license to share the lyrics for users on their site, Should google copy the "watermarked" lyrics and claim them as their own they would violate the license held by the first party. While they may not be violating the copyright of the first party they would however be violating the said license that they have paid for rights to use.

I would love to see a video that combines wonderful happy doggies and people and tasty educational content, for the duration of the video. Seeing a happy dog with a happy person always makes me super happy :3

Thank you :D

We very desperately need to update antitrust laws. Like, complete overhaul.

That’s fantastic 😂

For what it's worth, as a programmer, these watermarking approaches to text are fairly obvious. It's good that Genius took an organized approach to proving the conduct, but the information hiding here is fairly basic.

 If I was tasked with acquiring information by screen-scraping by a company like lyricsfind, I would refuse. But if for some reason I was not scrupulous to refuse, the most obvious thing would be to make an effort to take countermeasures against such fingerprinting. Automatically identifying and reducing any unicode sequences outside the most basic representations would be an obvious step, as well as folding away all unnecessary whitespace. Adding plausible deniability from there would not be terrible hard by introducing randomized algorithmic error in formatting, spelling, word choice, etc.

 Of course such efforts would only increase the legal risk, but they're not that hard to do.

 Basically, this was well done, but this shows more legal and organizational execution than technical excellence (and shows a severe lack of technical excellence by lyricsfind).

Who owns the copyright of transcriptions?

Google can just make Genius disappear from all searches if they want lol I can see some petty f@ckery happening in the future

Holy, I never expected A.V.E to get dropped there.

It's rare that a brand is descriptive. These folks are Genius.

Red Handed - BRILLIANT!!!!!!

I always thought they had a license to use Genius lyrics!

Leonard, I loved the channel already. When you mentioned MATTIAS WENDEL, my esteem for you went up exponentially. When you mentioned making a table out of walnut, I suspected you might be a brother from another mother. Let me know when you make your own bandsaw!!!

Who is Morris?....
MORRIS DAY AND THE TIME!

Interesting story it sounds like. When does it start?

Wait that's a green screen? The hell!!? 😧😧😧😧 This whole time I thought you were in your sweet ass garage My life is a lie!!

I want to know how this pans out, I believe that this should be treated as the original label/band owns the IP yes, but genius owns their transcription. I really hope they win.


However I'm sure that the next step, will be completely sanitizing the text fields, That would be searching for halfspace characters and if there's 2 of them, make a normal unicode space, all grave, apostrophe or other markers of that nature replace with standard apostrophes, then, normalize all carriage return characters, Completely sanitize formatting, of course The next move from Genius could be formatting in a way that would cause their sanatizing script to mangle the text, even going so far as potentially using a font that mismatches characters as run through a simple rot encryption. I've done something like this in years passed, so if you aren't using the same font, everything is 15 characters off, because the font itself is designed for my cypher so instead of A at the start it may be G (rot5) or maybe it's classical roman rot13) or you could use a more specific cypher to your cause maybe use a logical cypher that certain characters are rotated a different amount so A may be Rot5 but G may be rot 10 positions, so that you have certain characters are further off than others.


(note multirot requires that any time a character matches another so say that G is rot 10 so it would become Q, L also becomes Q so maybe you have the filter not move the L character. That way you only have one character value for each fontset character.)


you need a custom defined font set for that. however, if you copied it, wholesale font and all it's easy to prove that the unicode character doesn't match the displayed character, if you just copy the text content and put your own font to it, to normalize it, it mangles the content.

Shenaniganery must be my new favorite word..

Lawfulmasses makes joke about brandon stark.

New lawfirm opens up....

Lawwalkers and associates

Lawsuit is comming