4 ms·

Because for the life of them, Google can’t seem to create a single messaging system that works well. If they could simply make ONE system that works as a stand

by bborud 5y ago

Because for the life of them, Google can’t seem to create a single messaging system that works well. If they could simply make ONE system that works as a standalone product, has a decent API available to internal developers and external. That would be a start. But no. They have to make it part of something else. Always. And they always miss the target on usability.

tdeck 5y ago
That's not fair, gChat / Google Talk worked pretty well back in the day. I could even search my messages on mobile, which is something Hangouts never managed to offer in the many years after the migration.
- chii 5y ago
  and the reason it works so well is because gchat is based on xmpp which makes it more open to interoperability and access to pre-made libraries etc. And you can use any clients supporting xmpp. management at google must have decided that interoperability is a weakness and killed it.
  - bborud 5y ago
    Look at my comment above. Yes, at the time XMPP interoperability at the edges seemed like a good idea, but XMPP wasn't a good standard. No, the reason it worked well was because Google made their own backend that could actually scale and which wasn't based on XMPP at all. (I actually spent a lot of time reading the code and talking to the people who designed Talk while I was at Google, because that project had a lot of really good lessons in how to deal with challenges in distributed systems). The XMPP interop was nice, but it was just that: a client interface adapter. I can't think of a single XMPP backend that even comes close to working as well as the custom architecture Google used for Talk. Actually, XMPP is the worst kind of standard you can have. Something that superficially looks like a good solution, but doesn't lend itself to clear, correct and performant implementation. And at the same time is seen as THE standard for chat. So people won't be inclined to "start over". But if we're going to have chat interop, that's what we have to do. I'm not kidding. If XMPP was any good you would see companies that do chat products at scale prefer it, at least in the backend. Imagine all the time you can save. But they mostly don't. Ask people who develop chat systems why not. (Zoom is the only company that comes to mind, but chat isn't the main focus of that product, just an aside) Google killed XMPP interop because it was a pain in the ass and a technical obstacle. And then chat gradually became a closed affair because we, the developer community, just didn't come up with something that worked well enough to replace XMPP.
    - MattJ100 5y ago
      > Google killed XMPP interop because it was a pain in the ass and a technical obstacle. Yet Talk/XMPP was, tied with Hangouts, Google's longest-lived chat system to date (officially, anyway - last I knew a few weeks ago, it was still accepting XMPP client connections). XMPP has an open and diverse community, and I assure you there are many competent server developers involved in its development. Disclaimer, I'm heavily involved in XMPP, though I wasn't around for its conception. I've implemented it from both a client and server perspective. I'm not here to claim XMPP is perfect, but it's absolutely a good fit for its problem domain (an interoperable messaging standard). As a software engineer I've implemented many other protocols too. I assure you that all protocols have their quirks, XMPP included, but I've yet to see anything that would be a suitable replacement. I blame Google's culture as an organization for their inability to succeed at messaging. I blame financial incentives for large companies choosing to build silos rather than interoperate/federate. Whether that uses XMPP or not I really don't care - all the big players have had plenty of time to propose an alternative (or participate in XMPP development, as Google did for a period). I think only regulation can save us at this point.
  - rchaud 5y ago
    That explains why my 2008 BlackBerry had a Google-built Gtalk app. Almost no third parties were developing native apps for BBOS 6 at the time.
- bborud 5y ago
  Yes, Talk is clearly their best effort to date, but it wasn't great as a product. For instance it didn't have group chat that actually worked. Also, in retrospect, they did the wrong thing by choosing to offer XMPP as a public interoperability interface. I liked the idea of XMPP when it came out - or rather, the idea of having an IETF standard for chat. But XMPP had unnecessary problems due to poor design choices, probably informed by people who didn't have experience writing tidy and efficient server code. If you are going to design a network protocol you _have to_ involve people who know how to write server code in at least a handful of languages so that you know how to design stuff in mechanical sympathy with how you typically implement various aspects. Any standard written by people who are non-programmers or mediocre programmers will suffer. As a result of poor design choices, XMPP clients didn't turn out well. One example being that group chats were clumsy and ugly affairs in every single client. If they even supported group chat, which not all clients did. So I can understand Google eventually ditching interoperability. What they should have done is to just design their own protocol and open it, but at the time the choice was made to use XMPP a lot of people (myself included) really wanted it to succeed. (I actually blame XMPP for successful chat systems ending up being closed proprietary affairs. It shows how dangerous it is when people get infatuated with certain ideas and do not pay attention to what it takes to implement properly. When you have a poor standard that costs a lot to implement but which people kind of want to succeed it doesn't leave a lot of room for other standards to emerge. I know a lot of people will react negatively to this. I'd encourage people to try to implement an XMPP server well. Not just something that kinda works and pisses away a lot of resources: do it well. That's an unreasonable amount of work) As for Google: I'm not sure what was worse. Chat never being a goal in itself, but rather something tucked onto other services or the fact that to this day Google manages to make any form of chat or VC service confusing. Whenever I have to use a chat that isn't part of a calendar entry I do the same head-scratching exercise. Where do I go again? Which account will I end up on now (it seems to have a talent for ALWAYS picking the wrong account). At this point I'd rather not use Google for anything that has to do with communication. But it is hard to avoid. Unfortunately.

by bborud 5y ago

tdeck 5y ago

That's not fair, gChat / Google Talk worked pretty well back in the day. I could even search my messages on mobile, which is something Hangouts never managed to offer in the many years after the migration.

chii 5y ago
and the reason it works so well is because gchat is based on xmpp which makes it more open to interoperability and access to pre-made libraries etc. And you can use any clients supporting xmpp. management at google must have decided that interoperability is a weakness and killed it.
- bborud 5y ago
  Look at my comment above. Yes, at the time XMPP interoperability at the edges seemed like a good idea, but XMPP wasn't a good standard. No, the reason it worked well was because Google made their own backend that could actually scale and which wasn't based on XMPP at all. (I actually spent a lot of time reading the code and talking to the people who designed Talk while I was at Google, because that project had a lot of really good lessons in how to deal with challenges in distributed systems). The XMPP interop was nice, but it was just that: a client interface adapter. I can't think of a single XMPP backend that even comes close to working as well as the custom architecture Google used for Talk. Actually, XMPP is the worst kind of standard you can have. Something that superficially looks like a good solution, but doesn't lend itself to clear, correct and performant implementation. And at the same time is seen as THE standard for chat. So people won't be inclined to "start over". But if we're going to have chat interop, that's what we have to do. I'm not kidding. If XMPP was any good you would see companies that do chat products at scale prefer it, at least in the backend. Imagine all the time you can save. But they mostly don't. Ask people who develop chat systems why not. (Zoom is the only company that comes to mind, but chat isn't the main focus of that product, just an aside) Google killed XMPP interop because it was a pain in the ass and a technical obstacle. And then chat gradually became a closed affair because we, the developer community, just didn't come up with something that worked well enough to replace XMPP.
  - MattJ100 5y ago
    > Google killed XMPP interop because it was a pain in the ass and a technical obstacle. Yet Talk/XMPP was, tied with Hangouts, Google's longest-lived chat system to date (officially, anyway - last I knew a few weeks ago, it was still accepting XMPP client connections). XMPP has an open and diverse community, and I assure you there are many competent server developers involved in its development. Disclaimer, I'm heavily involved in XMPP, though I wasn't around for its conception. I've implemented it from both a client and server perspective. I'm not here to claim XMPP is perfect, but it's absolutely a good fit for its problem domain (an interoperable messaging standard). As a software engineer I've implemented many other protocols too. I assure you that all protocols have their quirks, XMPP included, but I've yet to see anything that would be a suitable replacement. I blame Google's culture as an organization for their inability to succeed at messaging. I blame financial incentives for large companies choosing to build silos rather than interoperate/federate. Whether that uses XMPP or not I really don't care - all the big players have had plenty of time to propose an alternative (or participate in XMPP development, as Google did for a period). I think only regulation can save us at this point.
- rchaud 5y ago
  That explains why my 2008 BlackBerry had a Google-built Gtalk app. Almost no third parties were developing native apps for BBOS 6 at the time.
bborud 5y ago
Yes, Talk is clearly their best effort to date, but it wasn't great as a product. For instance it didn't have group chat that actually worked. Also, in retrospect, they did the wrong thing by choosing to offer XMPP as a public interoperability interface. I liked the idea of XMPP when it came out - or rather, the idea of having an IETF standard for chat. But XMPP had unnecessary problems due to poor design choices, probably informed by people who didn't have experience writing tidy and efficient server code. If you are going to design a network protocol you _have to_ involve people who know how to write server code in at least a handful of languages so that you know how to design stuff in mechanical sympathy with how you typically implement various aspects. Any standard written by people who are non-programmers or mediocre programmers will suffer. As a result of poor design choices, XMPP clients didn't turn out well. One example being that group chats were clumsy and ugly affairs in every single client. If they even supported group chat, which not all clients did. So I can understand Google eventually ditching interoperability. What they should have done is to just design their own protocol and open it, but at the time the choice was made to use XMPP a lot of people (myself included) really wanted it to succeed. (I actually blame XMPP for successful chat systems ending up being closed proprietary affairs. It shows how dangerous it is when people get infatuated with certain ideas and do not pay attention to what it takes to implement properly. When you have a poor standard that costs a lot to implement but which people kind of want to succeed it doesn't leave a lot of room for other standards to emerge. I know a lot of people will react negatively to this. I'd encourage people to try to implement an XMPP server well. Not just something that kinda works and pisses away a lot of resources: do it well. That's an unreasonable amount of work) As for Google: I'm not sure what was worse. Chat never being a goal in itself, but rather something tucked onto other services or the fact that to this day Google manages to make any form of chat or VC service confusing. Whenever I have to use a chat that isn't part of a calendar entry I do the same head-scratching exercise. Where do I go again? Which account will I end up on now (it seems to have a talent for ALWAYS picking the wrong account). At this point I'd rather not use Google for anything that has to do with communication. But it is hard to avoid. Unfortunately.