Gee, it's almost as if that's the whole point of an ever-evolving SaaS platform.
A SaaS solution that claims to be private but won't provide the backend code to prove it. You don't find it at all suspicious that they claim releasing backend code would make it less secure? What kind of security product is not open for inspection? The same kind of "security" you get from Microsoft.
I imagine it probably is inspected, just not by the public. They probably do it themselves.
And they may have contracts with certain companies specializing in this sort of security that also inspect it.
And there's also the cybersecurity companies that test it whether they're contracted or not. At some companies, their entire job revolves around finding bugs (especially security bugs) in other companies' software.
Just because it's not on GitHub doesn't mean it's not a good product that hasn't been thoroughly tested.
Surely we're not gullible enough to accept "we inspected ourselves and determined we are secure and you should use our services"?
That's where the second and third paragraphs come in. Because other companies likely test it themselves, too.
They'll typically report security bugs privately and then, after X amount of months, publicly announce the bug. Doing it this way will, ideally, force the other company to patch the bug prior to the announcement. If not, they'll end up with a publicly known security bug that bad actors can now exploit. The announcement will also let the public (including companies) know to update their software.
Yes, and those other paragraphs are the same thing other proprietary companies do. Your opening paragraph is just absurd on the face of it because "inspected" does not mean "by themselves".
The second paragraph is literally speculation about something that might happen.
The third paragraph is about bug bounties, which every major software company does and which does not involve code inspection.
You just smokescreened and talked around the fact that your opening statement "it probably is inspected" is entirely unverifiable and non-credible even if true. I guess since you started that sentence with "I imagine" then it is technically true. You did imagine that.
I admittedly should've done more research before my first comment, but it does actually turn out that everything I said is true. Proton's technology was previously audited by Mozilla and is currently audited by SEC Consult and other companies regularly, and the audits are available for everyone to view. Additionally, they do have a bug bounty program. Also (and this is something I didn't mention), the ProtonVPN and Proton Mail apps are all open source.
Is that the backend code? It seems like they're talking about the apps, not backend code. The thing being discussed here is backend code.
Nearly all of Proton's stuff uses publicly verifiable client side encryption, so idk what all this is about
It's about the server-side code. If that's not an issue then someone needs to make the argument, not throw up smokescreens about the apps and frontend code.
You're right that the encryption needs to be verifiable on the client side, but then why not share the server side code?
I mean if they did, anyone could theoretically spin up an instance, which would be good, actually.
would be good, actually.
Good for us. Bad for business. I explained this in another comment too but Proton's idea of "open source" is simply to build trust in the security and privacy offered by the service. At least, as much as you can trust any SaaS.
but then why not share the server side code?
And to answer this... Well, business and practicality... One more than the other ofc unfortunately... Why would they take on the additional burden of making it self-hostable, make the backend fully open source, etc just to make competition for themselves? And that maintenance burden is huge btw, especially when the backend was probably never intended for self-hosting in the first place.
If Proton, as a company or foundation, didn't keep making the right decisions in terms of privacy and security, we might have had a reason to doubt their backend. But so far, there's been nothing. And steps like turning to a foundation-based model just inspires more trust. By using client-side encryption, even within the browser, they're trying to eliminate the need for trusting the closed source backend. Open sourcing the backend wouldn't improve trust in the service itself anyway since you can't verify that the code running in the backend is the same as the open sourced code. If you're concerned about data, they also offer exports in open formats for every service they offer.
Why wouldn't you trust them just because their backend is closed source? Ideologically, yeah I'd like them to open source absolutely everything. But as a service, whose income source is exclusively the service itself, how can it make sense for them to open source the backend when it cannot tangibly benefit their model of trust?
These are all good points and I have nothing to argue about with this comment. I really just wanted someone to answer the issue raised instead of changing the subject, and you've done that.
As for the linked comment, there are ways to verify that backend code is the same as open source. Not on a software level of course, but if you trust audits for logging practices presumably you can trust them for checking that the code base is the same.
Also you can verify that a web client is running the same code as open sourced, especially if it's a scripted client, since it would deliver code uncompiled. You can also check the signatures of binaries. Most people won't do this, but it only takes one security expert to check and discover that there's a discrepancy. If they then decompile it and find malware, that's the ballgame. Trust gone. There's a strong incentive for a premium service whose main selling point is privacy and transparency to never even flirt with that.
I agree that Proton has made themselves about as trustworthy as any private company can be, and maybe with the shift to foundation they can alter their model to not rely on being the singular operator. However, when you say "good for us, bad for business", that's the issue. The reason the fediverse works is that nobody can develop a monopoly on it. I mean, you've already said that ideally it should all be open source, so we agree on that too.
I understand that a closed backend isn't a deal breaker for a lot of people and that makes sense given the client side encryption. It's just that it is a potential problem in the longer term. It's an artefact of them having to exist in a capitalist context. Maybe they'll find a way through without succumbing to capitalist logic. I certainly hope they can.
In general, I agree with you. I would very much prefer if they did more open sourcing too. Just want to address some additional stuff.
especially if it's a scripted client, since it would deliver code uncompiled.
Unfortunately, this isn't really true anymore because of the necessity of minification. It introduces obscurity but is necessary for performance. But yes, the rest is correct, which is why I specified "web clients". You can verify the native clients, which is why native clients are so important imo. The concern of a hacked server serving a keylogging web client is unfortunately very real. Kind of makes it impossible to fully trust any SaaS at all.
if you trust audits for logging practices presumably you can trust them for checking that the code base is the same
The thing is, they already do public third party audits already. You can view their audit reports on their site. This is unlike companies like Google and Microsoft who conduct audits and keep the reports private. If you end up having to trust third party audits anyway, it doesn't help their model of trust since they do already do that in a transparent manner.
But yeah... stuff like the monopoly is kind of intentional. The exports are a mitigation, a huge one at that. Proton Mail exports are supported by services like FastMail, Proton Pass exports are supported by Bitwarden, etc. But in the end, the best case scenario would be some level of open sourcing. It's just that this "monopoly" is by design. For better or for worse, the fact that there is only one Proton is also good for Proton's model of trust tbh since the user doesn't have to wonder if the "instance" they're using is a good one for example. The fediverse model will not work for something that is so heavily based on trust. Proton wants to appeal to the general user, more than us folks... for better or for worse...
I hope they succeed too. I don't trust many companies. Proton has been one of the exceptions and I hope it stays that way...
The way I read it, they already (in the third paragraph of the blog post) had companies auditing their backend technology and (in the fourth paragraph) were starting to have companies audit their apps, too.
You realize that Microsoft code is inspected as well, even more heavily and regulated... and yet they still end up with major breaches. Security evolves through open source collaboration and inspection by experts that aren't being paid to say you're doing a good job.
You are making a lot good points... But is there any other practical solution?
Seems this is the best a normie on budget can get
They're not actually good points at all... Proton's open sourcing of the clients is for the purpose of trust in terms of security and privacy. The backend doesn't matter because the point is that the data is encrypted before it ever gets to the backend. The goal with Proton's open sourcing is not the ability to make it self-hostable. Sure, a lot of concerns are valid, but this isn't like Microsoft or Google. Nearly all of Proton is verifiably and provably secure. Well, at least as long as you trust the web clients being served are the ones whose code is publicly available. But again... You can't verify that with any SaaS. Such a risk is even present with self-hosting tbh. But that's another discussion.
You don’t find it at all suspicious that they claim releasing backend code would make it less secure? What kind of security product is not open for inspection?
No, because Proton has 3rd party audits all the time and they share the results openly.
Microsoft has third party audits all the time and say they're secure, and then you learn of new backdoors every 6 months. Audit companies are unreliable and paid to give good feedback while doing the least work possible.
Yeah because enterprises primarily use a ton of open source security tools...
ಠ_ಠ
Enterprises are using a plethora of open source tools at this point. They may still utilize closed source solutions, but they definitely have quite a bit of open source solutions tied in.
Releasing unfinished products and expect users to just make do while they launch the next product can't be the solution either.
Then it's a good thing all of their products are fully functional and working as advertised, I guess.
Sure, whatever you want to belief :)
Which bits are not functional? I'm using their email and calendar.. they aren't completely polished, but they're very usable.
Drive has no Linux client, Photos is extremely barebones and locks you basically in, as there is no export function.
Pass still has no proper SimpleLogin integration, no credit card support and UX wise is the browser extension pretty bad. Funny enough, years after launch you still can't auto fill on Reddit.
The only thing I don't like about Mail is that you still have to create reverse aliases through SimpleLogin. Better integration would be great.
Contacts still don't sync to you local mobile contacts. Which means you either do it manually or you have to keep two sets updated.
Calendar is good too, I've heard it has no offline support though. Although I haven't verified that.
Last thing I would like to see is notification support without Play Services.
Some of those things might be super unimportant to some, but for me it makes the use of their stuff unnecessary cumbersome. Especially if you consider that those are all Proton products and should work together well.
My by far biggest problem is their communication and general development speed though. Stuff like contact sync has been requested for 5(?) years now but there hasn't been so much as a "we're working on it".
It feels to me they come out with new products all the time, like the document editor now, without addressing the little things that would make their ecosystem great.
Anyway, long ramble. But I appreciate that you asked for more details without insulting me.
These are some excellent points, thank you for sharing rather than just giving a blanket "they bad" statement.
I believe Pass has integrated SimpleMail now, you can create aliases which forward to your email without setting up reverse aliases.
You can also add credit cards to Pass now, this was actually one of the things keeping me on Bitwarden for ages.
Can't say that their communication has improved though, all I can find on contact sync is that "Soon you'll be able to sync the contacts in your proton mail app to the default contact app on your mobile device" so if poor communication is your biggest problem then I can't fault you for avoiding them.
Drive has no Linux client,
I'm actually pretty sure they have one. I was doing a lot of exploring the last 2 days to make sure it was worth it to me to spend money on. And I landed on a downloads page on my Linux desktop that had a download link for drive for fedora, or debian. I can't find it on mobile (where I am now), but I'll look later on my PC and see if I can link it.
Was wrong, not sure what I found. Must have been the vpn or mail app.
Photos is extremely barebones and locks you basically in, as there is no export function
Interesting, thanks for the heads up. Hopefully it gets better later on, but for the moment I'm glad I made my own solution using a NAS, and a sync client.
Contacts still don't sync
This was one of my first concerns, I'm also annoyed by it.
Additionally, I was hoping their big "docs update" would also include spreadsheets, but hopefully soon.
Community has been begging for contacts for years...
It is getting tiring
That's a good list. Certainly a public feature/bug tracker would be nice. But those are pretty rare for corporate software..
Gee, it's almost as if that's the whole point of an ever-evolving SaaS platform.
A SaaS solution that claims to be private but won't provide the backend code to prove it. You don't find it at all suspicious that they claim releasing backend code would make it less secure? What kind of security product is not open for inspection? The same kind of "security" you get from Microsoft.
I imagine it probably is inspected, just not by the public. They probably do it themselves.
And they may have contracts with certain companies specializing in this sort of security that also inspect it.
And there's also the cybersecurity companies that test it whether they're contracted or not. At some companies, their entire job revolves around finding bugs (especially security bugs) in other companies' software.
Just because it's not on GitHub doesn't mean it's not a good product that hasn't been thoroughly tested.
Surely we're not gullible enough to accept "we inspected ourselves and determined we are secure and you should use our services"?
That's where the second and third paragraphs come in. Because other companies likely test it themselves, too.
They'll typically report security bugs privately and then, after X amount of months, publicly announce the bug. Doing it this way will, ideally, force the other company to patch the bug prior to the announcement. If not, they'll end up with a publicly known security bug that bad actors can now exploit. The announcement will also let the public (including companies) know to update their software.
Yes, and those other paragraphs are the same thing other proprietary companies do. Your opening paragraph is just absurd on the face of it because "inspected" does not mean "by themselves".
The second paragraph is literally speculation about something that might happen.
The third paragraph is about bug bounties, which every major software company does and which does not involve code inspection.
You just smokescreened and talked around the fact that your opening statement "it probably is inspected" is entirely unverifiable and non-credible even if true. I guess since you started that sentence with "I imagine" then it is technically true. You did imagine that.
I admittedly should've done more research before my first comment, but it does actually turn out that everything I said is true. Proton's technology was previously audited by Mozilla and is currently audited by SEC Consult and other companies regularly, and the audits are available for everyone to view. Additionally, they do have a bug bounty program. Also (and this is something I didn't mention), the ProtonVPN and Proton Mail apps are all open source.
Is that the backend code? It seems like they're talking about the apps, not backend code. The thing being discussed here is backend code.
Nearly all of Proton's stuff uses publicly verifiable client side encryption, so idk what all this is about
It's about the server-side code. If that's not an issue then someone needs to make the argument, not throw up smokescreens about the apps and frontend code.
You're right that the encryption needs to be verifiable on the client side, but then why not share the server side code?
I mean if they did, anyone could theoretically spin up an instance, which would be good, actually.
Good for us. Bad for business. I explained this in another comment too but Proton's idea of "open source" is simply to build trust in the security and privacy offered by the service. At least, as much as you can trust any SaaS.
And to answer this... Well, business and practicality... One more than the other ofc unfortunately... Why would they take on the additional burden of making it self-hostable, make the backend fully open source, etc just to make competition for themselves? And that maintenance burden is huge btw, especially when the backend was probably never intended for self-hosting in the first place.
If Proton, as a company or foundation, didn't keep making the right decisions in terms of privacy and security, we might have had a reason to doubt their backend. But so far, there's been nothing. And steps like turning to a foundation-based model just inspires more trust. By using client-side encryption, even within the browser, they're trying to eliminate the need for trusting the closed source backend. Open sourcing the backend wouldn't improve trust in the service itself anyway since you can't verify that the code running in the backend is the same as the open sourced code. If you're concerned about data, they also offer exports in open formats for every service they offer.
Why wouldn't you trust them just because their backend is closed source? Ideologically, yeah I'd like them to open source absolutely everything. But as a service, whose income source is exclusively the service itself, how can it make sense for them to open source the backend when it cannot tangibly benefit their model of trust?
My other comment regarding proton and trust: https://lemmy.world/comment/11003650
These are all good points and I have nothing to argue about with this comment. I really just wanted someone to answer the issue raised instead of changing the subject, and you've done that.
As for the linked comment, there are ways to verify that backend code is the same as open source. Not on a software level of course, but if you trust audits for logging practices presumably you can trust them for checking that the code base is the same.
Also you can verify that a web client is running the same code as open sourced, especially if it's a scripted client, since it would deliver code uncompiled. You can also check the signatures of binaries. Most people won't do this, but it only takes one security expert to check and discover that there's a discrepancy. If they then decompile it and find malware, that's the ballgame. Trust gone. There's a strong incentive for a premium service whose main selling point is privacy and transparency to never even flirt with that.
I agree that Proton has made themselves about as trustworthy as any private company can be, and maybe with the shift to foundation they can alter their model to not rely on being the singular operator. However, when you say "good for us, bad for business", that's the issue. The reason the fediverse works is that nobody can develop a monopoly on it. I mean, you've already said that ideally it should all be open source, so we agree on that too.
I understand that a closed backend isn't a deal breaker for a lot of people and that makes sense given the client side encryption. It's just that it is a potential problem in the longer term. It's an artefact of them having to exist in a capitalist context. Maybe they'll find a way through without succumbing to capitalist logic. I certainly hope they can.
In general, I agree with you. I would very much prefer if they did more open sourcing too. Just want to address some additional stuff.
Unfortunately, this isn't really true anymore because of the necessity of minification. It introduces obscurity but is necessary for performance. But yes, the rest is correct, which is why I specified "web clients". You can verify the native clients, which is why native clients are so important imo. The concern of a hacked server serving a keylogging web client is unfortunately very real. Kind of makes it impossible to fully trust any SaaS at all.
The thing is, they already do public third party audits already. You can view their audit reports on their site. This is unlike companies like Google and Microsoft who conduct audits and keep the reports private. If you end up having to trust third party audits anyway, it doesn't help their model of trust since they do already do that in a transparent manner.
But yeah... stuff like the monopoly is kind of intentional. The exports are a mitigation, a huge one at that. Proton Mail exports are supported by services like FastMail, Proton Pass exports are supported by Bitwarden, etc. But in the end, the best case scenario would be some level of open sourcing. It's just that this "monopoly" is by design. For better or for worse, the fact that there is only one Proton is also good for Proton's model of trust tbh since the user doesn't have to wonder if the "instance" they're using is a good one for example. The fediverse model will not work for something that is so heavily based on trust. Proton wants to appeal to the general user, more than us folks... for better or for worse...
I hope they succeed too. I don't trust many companies. Proton has been one of the exceptions and I hope it stays that way...
The way I read it, they already (in the third paragraph of the blog post) had companies auditing their backend technology and (in the fourth paragraph) were starting to have companies audit their apps, too.
You realize that Microsoft code is inspected as well, even more heavily and regulated... and yet they still end up with major breaches. Security evolves through open source collaboration and inspection by experts that aren't being paid to say you're doing a good job.
You are making a lot good points... But is there any other practical solution?
Seems this is the best a normie on budget can get
They're not actually good points at all... Proton's open sourcing of the clients is for the purpose of trust in terms of security and privacy. The backend doesn't matter because the point is that the data is encrypted before it ever gets to the backend. The goal with Proton's open sourcing is not the ability to make it self-hostable. Sure, a lot of concerns are valid, but this isn't like Microsoft or Google. Nearly all of Proton is verifiably and provably secure. Well, at least as long as you trust the web clients being served are the ones whose code is publicly available. But again... You can't verify that with any SaaS. Such a risk is even present with self-hosting tbh. But that's another discussion.
No, because Proton has 3rd party audits all the time and they share the results openly.
Microsoft has third party audits all the time and say they're secure, and then you learn of new backdoors every 6 months. Audit companies are unreliable and paid to give good feedback while doing the least work possible.
Yeah because enterprises primarily use a ton of open source security tools...
ಠ_ಠ
Enterprises are using a plethora of open source tools at this point. They may still utilize closed source solutions, but they definitely have quite a bit of open source solutions tied in.
Releasing unfinished products and expect users to just make do while they launch the next product can't be the solution either.
Then it's a good thing all of their products are fully functional and working as advertised, I guess.
Sure, whatever you want to belief :)
Which bits are not functional? I'm using their email and calendar.. they aren't completely polished, but they're very usable.
Drive has no Linux client, Photos is extremely barebones and locks you basically in, as there is no export function.
Pass still has no proper SimpleLogin integration, no credit card support and UX wise is the browser extension pretty bad. Funny enough, years after launch you still can't auto fill on Reddit.
The only thing I don't like about Mail is that you still have to create reverse aliases through SimpleLogin. Better integration would be great.
Contacts still don't sync to you local mobile contacts. Which means you either do it manually or you have to keep two sets updated.
Calendar is good too, I've heard it has no offline support though. Although I haven't verified that.
Last thing I would like to see is notification support without Play Services.
Some of those things might be super unimportant to some, but for me it makes the use of their stuff unnecessary cumbersome. Especially if you consider that those are all Proton products and should work together well.
My by far biggest problem is their communication and general development speed though. Stuff like contact sync has been requested for 5(?) years now but there hasn't been so much as a "we're working on it".
It feels to me they come out with new products all the time, like the document editor now, without addressing the little things that would make their ecosystem great.
Anyway, long ramble. But I appreciate that you asked for more details without insulting me.
These are some excellent points, thank you for sharing rather than just giving a blanket "they bad" statement.
I believe Pass has integrated SimpleMail now, you can create aliases which forward to your email without setting up reverse aliases.
You can also add credit cards to Pass now, this was actually one of the things keeping me on Bitwarden for ages.
Can't say that their communication has improved though, all I can find on contact sync is that "Soon you'll be able to sync the contacts in your proton mail app to the default contact app on your mobile device" so if poor communication is your biggest problem then I can't fault you for avoiding them.
I'm actually pretty sure they have one. I was doing a lot of exploring the last 2 days to make sure it was worth it to me to spend money on. And I landed on a downloads page on my Linux desktop that had a download link for drive for fedora, or debian. I can't find it on mobile (where I am now), but I'll look later on my PC and see if I can link it.Was wrong, not sure what I found. Must have been the vpn or mail app.
Interesting, thanks for the heads up. Hopefully it gets better later on, but for the moment I'm glad I made my own solution using a NAS, and a sync client.
This was one of my first concerns, I'm also annoyed by it.
Additionally, I was hoping their big "docs update" would also include spreadsheets, but hopefully soon.
Community has been begging for contacts for years...
It is getting tiring
That's a good list. Certainly a public feature/bug tracker would be nice. But those are pretty rare for corporate software..