Yeah, it creates an adversarial relationship which isn't particularly pleasant. E.g. they have no incentive to build functionality to help integrate, nor do they have the same incentive to provide stable integration points. In fact, it sounds like stable integration points are borderline actively dissuaded as they would limit agility of Signal and relegate it to irrelevance.

As much as I disagree with the underlying values, I haven't carved out 40 million monthly active users from a crowded space, so maybe it really is the best chance.

Nobody needs to absolutely prevent third-party clients; they just need to create enough friction to prevent serious investment in those clients. At least in Signal's case, there's a clear rationale for why they're doing that.

And yet whatsapp, discord and slack are all doing pretty well at it.

I kinda feel like that's more due to lack of interest than anything else, at least on the part of WhatsApp, which is restricted enough feature-wise that it'd be somewhat feasible to reverse engineer and build a client.

If I think about something like Slack or Discord, though, they surface area of those things are just huge. So many protocol details to work out, and so many features to duplicate for a third-party client. And they're such moving targets, too, with features added at a regular cadence (unlike AIM, ICQ, etc. back then, though even then Gaim/Pidgin didn't have voice & video support for a long time).

I also think the third-party client drive for AIM and ICQ was driven by people who wanted to use it on Linux but had no way to do so at all. The clients were Windows- (and Mac-?) only, WINE was a pain to use back then, and there were no web clients initially. But with Slack and Discord the impetus just isn't there. At least with Slack (I'm not super familiar with Discord), Windows, Mac, Linux, Android, and iOS are all officially supported, and there's always the web client if you don't like the Electron-based desktop clients.

On top of that, Slack's API is rich enough to enable things like a Matrix bridge, so, again, probably just not enough interest to do something truly standalone.

The only thing I'm somewhat surprised about is that no one has tried to build a third-party standalone WhatsApp client that doesn't require being tethered to the mobile client. But perhaps Facebook/WhatsApp has done a better job at obfuscating and hiding their protocol than AOL and Mirabilis did long ago. Or maybe the interest just isn't strong enough to warrant the effort.

Interesting, what makes you think the surface area of Slack is larger than that of WhatsApp, for example?

I think the latter hides a lot of complexity behind a simple UI, most notably all the E2E encryption logic and the purely client-side message storage, with an optional backup system and even its own built-in remote access server (for the web client).

Compared to that, Slack is mostly a HTML renderer as far as I can tell, with most of the complexity server-side (e.g. complex notification rules, multi-client support etc.)

I suppose the poster above you is referring to user experience - WhatsApp doesn't even have reactions, never mind the huge list of plugins that slack has. The only thing in favour of WhatsApp as a user is the encryption, which AFAIK slack just doesn't bother with (maybe that changed now).

I‘m still not convinced that all of these user-facing features necessarily translate into a larger client codebase.

My guess is that WhatsApp (and Signal, Matrix etc.) client logic is much more complex, since so many things that are fairly easy to do server-side need to be reliably performed in the client instead once E2E encryption is introduced.

Actually I think it's because WhatsApp bans you if it detects an unofficial client. It's not worth the risk.

I would love to have a anonymous WhatsApp client. Even if the features set would include only reading and writing.

Same here. I'm literally forced to use it with no more than a dozen contacts among close friends and part of the family, and despite the very short list of people, it's absolutely impossible to convince anyone of them to use anything else. I would so welcome a FOSS client.

The official WhatsApp client has an excellent UX. There's simply no reason to use another.

For Signal it's not the case. The client is slow, has many quirks and lack features.

If Moxie allowed third-party clients, it would be a win-win situation. Over time users will get a usable client, and the core Signal team would be able to concentrate on their Crypto business.

I completely agree that Signal could very much use a third-party client, especially on the desktop. That being said if they want to push their cryptocrap they have to keep a strong control on the clients because third parties are likely to just ignore that functionality altogether, hampering adoption.

I'm not saying that's the main reason for them to exert such a tight control on the Signal ecosystem but it shows how damaging that move has been: now you can't really exclude an ulterior motive for them refusing to integrate with third parties. It damaged the trust I had in the project.

The signal desktop app is severely neglected. The UX is terrible and users have been begging for it to be able to interact with regular SMS for years.

Is everyone using another WhatsApp client than me? If I would rate all chat clients I have used I would put it in the bottom 2, only Skype has a worse client.

there are matrix bridges for all three of those

https://github.com/tulir/mautrix-whatsapp

https://github.com/Half-Shot/matrix-appservice-discord

https://github.com/matrix-org/matrix-appservice-slack

The whatsapp bridge is based on the whatsapp-web api, which is non-functional if your phone is off. The phone part is what needs reverse engineering and replacing.

The Discord and Slack bridges use the respective platforms' official APIs, fwiw

I can't speak to the others but judging by the amount of spam I get on Discord, they certainly aren't.

Ripcord exists, which is a slack AND discord client made by a third party.

big difference between using the name Signal and writing a third party client.