FairScan developer accuses Ente of copying code without attribution

2 Likes

Oof… That’s rough

This is pretty damning, especially considering Ente has since acknowledged that parts of its original scanner were adapted from that code. I’m not an expert on this stuff, but I’m not sure if that makes for a GPL violation, but I’d say he has every reason to question how this was handled. At the very least he should get a proper explanation.

1 Like

This sucks. Fairscan is great, but is it really that hard to just give credit where it’s due? I mean, by the look of it, it’s not even a “look, I rewrote this GPL thing, now it can be MIT!”, situation, so… why would Ente not give proper attribution?

6 Likes

The more I learn about Ente the more questions I have about how they actually run their business. I’m not even trying to keep up but somehow there’s always something new like this coming out.

First all of this weird narrative like “is Ente European company?” (Ente means duck in German and they have a duck logo), then “is Ente American company?” (registration in US) while Ente is a company mostly operating from India. Then came the whole Proton drama, including accusations around Google Ads. And considering that Ente’s CEO used to work at Google and they seem to have as many (or more) people doing marketing as actually developing the product, it’s pretty wild that nobody knew how Google ads work. Then I found out about Ente Privacy Pack controversy through Reddit, if you missed it: Reddit

And now this situation with the code. In short: according to the developer, Ente contacted him out of the blue thanking him for FairScan and saying they were using his segmentation model in Ente Locker but on their blog post they claimed they had written the code around it “from scratch”. When he noticed, he compared the code, he found almost direct translations of his Kotlin implementations into Rust with the same structure, algorithms, function and variable names, and even hard coded values. When confronted, Ente dismissed the similarities as standard methods, then rewrote over 6,000 lines of code within three days, leaving developer’s questions unanswered.

If these allegations about passing off someone elses code as your own are true, that’s a pretty serious problem.

At some point you have to wonder how many weird coincidences a company can have before people start noticing a pattern. Maybe the marketing team could spend a little less time managing the narrative and a little more time making sure there’s nothing to manage. Repeatedly finding yourself in situations where people have to question your transparency is probably not the best growth strategy for Ente.

3 Likes

All these issues are inconsequential.

I feel for the developer. This probably felt soul crushing.

This sentiment is the exact opposite of what open source developers should be feeling:

What happened is something I wasn’t ready for. I put a lot of myself into FairScan, perhaps too much: seeing my work dismissed as if it had no value is very hard not to take personally. Today, this makes me question my motivation to contribute to open-source software.

Hoping a resolution is found that restores faith in the open-source community.

5 Likes

How very, very corporate of them (Ente).

There is a bit of a misunderstanding here.

The code screenshots shared on the FairScan blog are from when our engineer had originally reached out to them, after crediting them in our blog, newsletter, and socials. When FairScan expressed discomfort, our engineer felt (perhaps wrongly so) that attribution was not the solution being asked for, and rewrote the code completely.

We have since added attribution in our source code to acknowledge provenance, in good faith.

We are also in conversation with FairScan to resolve this matter to their satisfaction - we have even offered to delete the feature completely.

FYI - the email that our engineer originally sent to FairScan.

9 Likes

Sorry, but that’s not really what the FairScan blog describes.

You keep mentioning that FairScan was credited in your blog but that’s not the same as crediting him in the actual GitHub repository. People looking at the code wouldn’t see any of your blog posts. If the code came from FairScan, that should have been acknowledged where the code itself lives.

And the issue isn’t just about GitHub attribution. You stated at first that image processing code was written by Ente from scratch, while FairScan showed pretty clear similarities between the two implementations.

Saying you rewrote the code doesn’t really address the original issue either. What about the code that was already shipped in Locker 1.0.9? Was it compliant with the GPL or not?

You also say you’re working with FairScan to resolve this but his latest update says he still hadn’t received a response to the email he sent on September 21.

Adding attribution to GitHub repository now is a good step but it doesn’t change the fact that it wasn’t there before. And offering to remove the feature doesn’t explain how this happened in the first place.

If this really was just a misunderstanding then why not address the actual points FairScan raised?

3 Likes

We were the ones who reached out to FairScan in the first place, which would be a strange thing to do if our intention was to steal code. We are in conversation with FairScan and are working with them to reach an amicable resolution.

Given that your entire posting history here seems to focus on Ente and its supposed shortcomings, I don’t think there’s anything I could share that would change your mind. I’m therefore not going to litigate this matter any further with you.

10 Likes

Thanks for taking the time to come and try to clear things up a bit. Personally, as long as we reach a solution where FairScan and Ente are both happy, I think that’s all that really matters. Misunderstandings certainly can happen. People make software fun, but also a little messy sometimes.

4 Likes

If I wanted an answer just for myself, I would’ve sent you a DM. But this is a public discussion and I think it’s valuable for everyone to be able to read what a company has to say about issues like this.

I do comment on Ente related threads quite often. I just find it incredibly unique (for lack of a better word) how many weird situations keep coming up around one company. You don’t really see this kind of thing happening with other companies to the same extent.

Since this is a public discussion I think those questions are still worth asking.

1 Like

I do realize there can be licensing requirements and common practices to consider. But still:

I’m subscribed to the Ente blog, and when I read the post about the Locker update, the first thing I did was visit the FairScan website to learn more. I’m not sure if that was the first version of the post or a later one. But from a user’s perspective the post did bring FairScan to my attention, and felt like a good example of devs sharing their audience with another project

Again, the devs’ perspective may be different. I’m describing my experience as a reader. Hope both projects work things out

4 Likes

Never heard of FairScan, but now I can’t wait to try it.

If all the developer required was an honest acknowledgement of the use of their code, I don’t understand why Ente wouldn’t be forthright about that. It doesn’t make them less innovative as a company. It never occurred to me that closed-source companies could just use open-source code and never acknowledge it. That is messed up.

I also hate that users and small developers need to come out publicly in order for companies to respond to controversies that they were given ample opportunity to deal with privately.

2 Likes

I agree with @Luke here. Even if they were an outspoken Ente critic, valid points need to be addressed, regardless of if their mind changes by this.
That is not to say you don’t, I mean that in general.
No need to treat it as if it would only benefit @Luke if the discussion goes on.

3 Likes

As I specified in the post, I don’t see anything constructive coming out from a discussion specifically with this one commenter. Happy to engage with others.

1 Like

He updated its post (this is only a snippet, please read the full post):

On October 21, Ente added the rust/crates/ml/README.md file to their repository: this file explicitly mentions FairScan.[…]

On October 24, Ente replied to my September 21 email, explaining that they hadn’t understood my emails and apologizing for how the situation had unfolded.


Maybe it is indeed a misunderstanding. I believe FairScan’s creator is French while Ente has location in the USA, India and the Netherlands. So, except for some peoples working at Ente in the US, all the people concerned aren’t necessarily native English speakers. Even if it’s fairly easy nowadays to translate everything in most languages, errors can still happen.

Also, no employee seems to have legal nor public relation formation as what can be seen in their about page. It might simply be a lack of license knowledge or of communication.

Not taking party, just adding elements to the discussion.

The language argument is pretty weak. Ente is a company, it’s fair to expect them to know how to communicate in English. Their apps are in English, right?

But the problem started long before that email. Ente had already failed to properly credit the original author on GitHub (before he had to call them out publicly just to get proper credit). GitHub repo is exactly where people look to see what a company actually built itself and what it took from someone else’s work. Proper attribution isn’t some minor detail you can just brush off. It matters because there’s a difference between developing something yourself and building on someone else’s work.

So let’s not distract from the actual issue by bringing up native languages. Ente is a company, not a bunch of random people who are just starting to learn how open source works.

Yes, I can see FairScan updated his post today. You didn’t include all he had to say though:

Blockquote On October 24, Ente replied to my September 21 email, explaining that they hadn’t understood my emails and apologizing for how the situation had unfolded.
Ente’s response frames the situation as a misunderstanding, which I find very difficult to reconcile with the sequence of events. It does not address the broader concerns I raised about how Ente handled FairScan’s code, nor does it acknowledge any specific mistake in how they handled the situation since September 16.

He said himself the concerns he raised still haven’t been addressed. I pointed them out in my earlier comment too and they still refuse to address them. That says a lot.

2 Likes

Blockquote lwb03dcEnte
As I specified in the post, I don’t see anything constructive coming out from a discussion specifically with this one commenter. Happy to engage with others.

@anon59491642 feel free to copy and paste my comment if you want. Maybe they’ll address those points for you. Or wait, maybe they won’t, since according to FairScan they haven’t even addressed them to him either :distorted_face:

1 Like