What Slack might learn from its Open Source alternative | Hacker News Reader

What Slack might learn from its Open Source alternative

mattermost.org

219 pointscaio198294 comments

Is this mattermost's hosted offering ? https://about.mattermost.com/pricing/

they need to fix their feature set for pricing. For example, Google Apps Signon is pretty basic - but it is only available at the "Enterprise E20" level... where you have to actually call for pricing.

It's really arrogant to claim that a pricing model must be broken just because you don't approve of the way the features are segmented.

I'm not going to claim that it's broken, but it does seem a little odd in that the cost to customers is not aligned with cost to the hosting provider.

Slack's free tier only allows you to keep a history of something like 500 messages, which makes sense because storage and indexing of the message history are what cost money for the provider.

This product let's you store all you want for free (seems like a bad idea to me), while things with essentially no cost per user once they have been developed (like google auth) are the things they make you pay for.

You are on Hacker News, yet you think that that's odd? It's a common wisdom among entrepreneurs that you price your product according to what people find valuable, not according to what it costs you. For example https://www.smashingmagazine.com/2011/09/youre-pricing-it-wr...

Summary: for many services, the things that seem like they're expensive ("storage and indexing") don't cost much. The things that seem like they should be free (features) are expensive, even if there's no variable cost.

Details: Regarding "storage and indexing of the message history are what cost money for the provider," this is often not the case. Although at extreme scale (like any AWS product), it's absolutely true, for a typical Web app, the difference between, say, 5 and 50 MB of chat history is not what costs money. 2 things do:

1. Paying developers (that is, adding features). As you note, they have no cost per user, but they have a huge onetime upfront cost. Just because a feature doesn't have a variable cost doesn't mean that the cost isn't real and isn't distributed among n users[1]. Ideally, that cost should be covered by those receiving value from it.

2. Support and operations. Depending on scale, quality, and so forth, assume 10-30% of your payments go towards answering emails and keeping the service operating. Offering a feature like Google Auth does have a variable cost when, say, 30% of those who use it are going to ask a question, and answering that question well is going to cost $25 in time. Sure, a service could say "We don't offer support for free plans," but some services rightly don't want anyone to get stuck, even if they aren't paying. It's also harder to do thoughtfully than it might seem, since many customers who will eventually pay ask questions before they're receiving enough value to upgrade.

[1]: Again, a few services operate at such large volume that the implementation and support cost is a minuscule percentage of revenue. That's relatively few Internet services, though, mostly very high volume hosting. For other products, even all but the largest PaaS/IaaS offerings, the items above are substantial costs. Some services which would otherwise be lower-volume use this as a differentiator ("Everyone gets all features"), and that's great, but there's no reason to expect it universally.

Slack lets you keep a history of 10,000 messages, and they're retained even if you're on the free tier. You just can't see them.

S3 is a cent a gigabyte at scale, three cents when you're starting. sc1 drives on EBS are two and a half cents. Storage is a cheap, solved problem.

You make people pay for what's important to them.

Same reason a 32GB iphone costs hundreds more for the next level of storage. And why 64gb is pretty much the largest offering. 64gb isn't enough? Pay for more icloud storage. Yet you can buy a 128gb sd card for $30.

I didn't say it was broken. But if we are addressing the startup set on HN.. Then I'm in the market to buy their service. I'm not a competitor, nor do I have a vested interest.

I get Hipchat for free because I'm a bitbucket customer ($1per month) and I get Google Auth as built in. Most startups are using Google Apps.. But very few would use LDAP (which is indeed available in the 20$ tier)

So yes - I believe for the demographic I represent, their pricing needs to be fixed.

So their pricing isn't broken yet you repeat your claim that it "needs to be fixed"? How is this not pedantry?

Look, I understand your desire to pay less for a service, but insisting that pricing needs to be fixed simply so that you pay less is still absurd.

>>I didn't say it was broken.

You said "they need to fix their feature set for pricing."

Not "improve."

"Fix."

That's the word you used. And it directly implies that the subject it is referring to (the pricing) is broken.

feature set - not pricing. Am I making a mistake with my english ?

I'm willing to pay their 20$ thing - but the features they include in that price is weird for the segment on HN. And for the features people like me would really pay for... there is no pricing (so I dont know whether its broken or not).

I dont want them to lower the pricing - I want them to rearrange it... or atleast talk about it. I dont know how much to pay to get the Google Apps feature.

You want to pay the 20$ thing but you want additional features from the more expensive call-us tier. That is fundamentally a complaint about their pricing.

The raison d'etre of pricing tiers is to charge a different price to different customers. Having to pay extra to get extra features is the point, not a bug that needs fixing.

Hi @sandGorgon, hi @gizmo

Mattermost team here. Thanks for your feedback,

While pricing is set that way for a purpose, we are always looking for ways to improve and we're keenly interested in what everyone has to say.

Our focus for Enterprise has been large organizations with sophisticated needs and it's important for us to have conversations and demonstrations with the people doing the evaluation so they fully understand what we're offering.

That said, there are people who use our free Team Edition or have read through our docs that just want a "buy" button for E20.

I'll bring this up and see what we can do.

PS: Regarding the word "fix", my personal view is that it's okay for a commercial product, because people are buying it as a customer. Enterprise Edition is a commercial product.

That said, many open source communities are adopting a principle of "kindness" and having feedback phrased positively where possible. People work hard to make open source software and give it away, and if conversations started on a positive note... well, why not?

hi, thanks for replying.

So at no point am I implying that you should release it for free. But I do believe there are some features that should ideally be in the 20$ tier rather than the "call-us" tier.

Now this may completely be how your demographic falls, but generally speaking a small company buying 2-10 licenses would usually not have LDAP... but most likely will have Google Apps. And that's pretty much the gist of my comment.

I'm not very sure why people have caught on to the word "fix" - but IMHO i used it in the way that most people talk on these forums ("your website font needs to be fixed", "the copy needs to be fixed"). So I'm not sure if the word was offensive, but it was not intended to be.

Also wanted to mention that I was looking at your offering as a customer with a genuine interest to buy. We use Hipchat and their mobile app needs to be "fixed".. so I was in the market for alternatives.

P.S. also didnt realize you were YC. That usually doesnt happen.

Thanks @sandGorgon,

Definitely hear your feedback on the custom tier, we'll try to figure something out this next week.

I also agree about Google apps being a good choice in a self serve tier; custom LDAP is more a true enterprise feature. You might be able to put some limits into the self serve Google apps version, too. I'd just like a user with small numbers of users to be able to sign up directly and use the product. (I've been a fan of the product for a while and will probably contact you in a month for enterprise)

Which communities don't approve of the word "fix" in describing suggestions?

I'm all for criticism being constructive and polite, but in my experience fix is a neutral and common word in design and development discussions.

Yeah, that comment about "fix" being negative was really rather passive aggressive.

This person has clarified several times that English is not their first language and they're making a positive suggestion of how you can better target small startups yet you're endorsing the criticism of their use of a single word. Lost respect for Mattermost (and HN) from this whole thread.

Tangential note - The data-lockin is exactly why I don't use Slack or that litany of SaaS solutions out there. Even with Oracle or MS SQL Server (or even Azure on-premises), if I needed to I could bulk copy my data out.

Open source + charging the enterprise a lot for customized solutions and/or paid support is a fantastic model and I wholeheartedly stand with your model.

(N.b., marketing strategy -- capitalize off the fact that everyone's data is being tapped now - e.g. Dianne Feinstein, in charge of the Senate Intelligence Committee's, had her staffers machines hacked - Apple is 100% hackable too (see: the San Bernadino FBI debacle) and offer secure comms at Blackberry Enterprise licensing pricing. I'd estimate > 95% of F500 is MS based/LDAP based on my experience. They're used to paying tons of money for Remote Desktop/AppX/Citrix/Exchange/whatever seats. You could easily get away with $60/user/seat + install fees + 50% for platinum support.)

Zulip is also open source and an alternative. It shares some of the same features as Mattermost. Mattermost seems to compete more directly with Slack; Zulip is more like IRC++. (For those from the MIT or CMU world, it's basically Zephyr with a web interface and a pile of features on top.)

Full disclosure: my company has been using Zulip since before it was F/OSS.

IRC die-hard here. Zulip is the only one of these Slack-like services I could ever see coming close to replacing IRC for me.

Can you explain how Zulip is more like a better IRC and Mattermost like Slack?

I'm a reluctant and weak (opposite of power) user of IRC and Slack and though I understand their great architectural differences, as a user they feel very similar. Surely I'm missing something.

I would define it more as Source Open than Open Source. Their Source is under Afero License but they only use the Acronym followed by a printout of the Apache License trying to trick people into thinkink it's Apache. And their binary MIT License has the addition that you aren't allowed to hide modify any Product or Trademark indentifications. That contradicts the MIT License!

Admittedly, their licensing scheme is a bit strange, but it's very clearly F/OSS. Even in the most restrictive case, it's still the AGPL with additional exemptions.

I'm not sure that the portion under AGPL actually has exemptions. They say in their LICENSE.txt file:

  We promise that we will not enforce the copyleft provisions
  in AGPL v3.0 against you if your application (a) does not
  link to the Mattermost Platform directly, but exclusively
  uses the Mattermost Admin Tools and Configuration Files, and
  (b) you have not modified, added to or adapted the source code
  of Mattermost in a way that results in the creation of a “modified
  version” or “work based on” Mattermost as these terms are defined
  in the AGPL v3.0 license.

Note that the two conditions necessary to receive the exemption are connected by "and", not "or". The second condition is that you not create a modified work or a work based upon the program as defined in AGPL v3.0.

But by the terms of AGPL itself its copyleft provisions are only applicable to modified versions, so it doesn't appear that there are any circumstances where Mattermost's exemption will actually exempt you from anything.

As a free software lover for me Affero GPL > Apache License for web-hosted tools any day!

Then why do they need to trick users/devs into thinking it is Apache?

One license makes networking applications sources available if modified, the other doesn't as far as I'm aware. Not sure why it has 3 "licenses" they really should figure that out. I think they should just say they're AGPL if anything and make any exception a paid license that turns Mattermost into MIT or Apache license or something? But having two licenses is fine, or one license and exemptions, but three and an exemption that doesn't really make sense... Sheesh.

Mattermost team here,

Highly appreciate your feedback.

Our intention is that 99% of people use Mattermost Team Edition under MIT.

You're right about that trademark section under MIT, that is weird, we'll remove it in the next release.

For the 1% of people who want to modify the core /platform, we're just asking they share back to the community.

Here's a summary of our licensing and philosophy: https://www.mattermost.org/licensing/

Well, mattermost is not entirely open source. Last I checked their LDAP feature was enterprise only and we went with Rocket.Chat which had all features open source.

If anyone wants to integrate Mattermost with their LDAP server, another way is to let Mattermost use your local GitLab to authenticate and let GitLab use LDAP. That's what we do. This should work out of the box with GitLab's omnibus package.

The project is open source, what you're complaining about is that the LDAP feature isn't free. You have to try to understand that they can't pour in thousands of hours across a development team without making money. How are their developers gonna get paid?

The golden nugget with these kind of projects, from these kind open source-minded companies, is that being a paying customer contributes greatly to a better and healthier product with regular feature additions. Compared to the more frequently found companies that is closed source, commerical-only and only provide bugfixes I think that it is perfectly fine that the LDAP-feature is behind a paywall.

> these kind open source-minded companies

Let's not fool ourselves, there is no such thing in these companies and products that appeared during the last few years. There is no free/open-source spirit, it's open-washing and the model usually get more closed and more closed year after year (going through an incredible variety of licenses, market segmentation and other marketing & sales tricks), following the evolution of the owners' greed.

It's a regular business that is freemium-minded, period.

Of Slack clones, Mattermost tries to be feature complete but is actually incredibly buggy. Let's Chat is strong, but is simply lacking important features. RocketChat is the only one that does a reasonable job being feature complete and not being full of bugs. It's actually usable...

Hi @chriscappuccio,

Could you share an example of what you went wrong?

The Mattermost server has around 59% automated test coverage, our community does over 120 hours of manual testing prior to each monthly release, and we're constantly striving to improve.

Here's a recent review of Mattermost's quality levels, from someone who left Mattermost and then returned, concluding Mattermost was the leading solution: http://www.akitaonrails.com/2016/08/13/choosing-mattermost-o...

If your experience is different, we're highly interested in knowing how we can improve.

We already installed RocketChat and test drived it in the our server room. It is really so many bugs and it looks very unstable. We are not understand that Why RocketChat has many bugs(upload attachment, video plug-in, etc..)

So far, we ranked #1 is Mattermost and other is Julip.

We've been using a self-hosted version of Mattermost for several months and the only complaint is that you can't switch off conversion of smileys to emoji. There are few minor nitpicks around email notifications and peers status display, but otherwise it's a solid no-nonsense alternative to Slack. Well done, guys!

Thanks @eps! Glad to hear!

Did you want to turn off smiley-to-emoji conversion for your own view, or on how others view your messages, or as an admin feature for the server?

There's an upvote and discussion thread on this, would welcome your input: https://mattermost.uservoice.com/forums/306457-general/sugge...

2nd and 3rd, please.

We are 40+ years olds, coming from the text terminals era, so emojis is really just some cutesy-girly thing that teenagers use :)

Mattermost is bundled with GitLab. We use both at our company and it works great.

That's cool. Are they integrated well?

Mattermost can use GitLab for authentication, which is nice. Other than that, you can configure webhooks in GitLab's git repositories to push messages to a Mattermost channel; for example, push every tag created so colleagues can see a new version of one of your libraries/products was released.

There's work in progress with Cog for increasingly powerful Mattermost-GitLab integration: https://forum.mattermost.org/t/help-wanted-cog-integration-f...

Full markdown in messages is really useful, we make heavy use of it. Its integration with GitLab could use some polish though, images in tickets don't appear in channel notifications for example

Again, a product that advertises being built on Go as if that's a feature. I have nothing against Go, but that's weird -- no one cares what Slack is written in.

Personally I care about the security of the server (is it written in a language that historically has a good or poor track record), and I care a great deal about the client.

The Slack client regularly pauses for me while I am typing, randomly refreshes, and generally feels sluggish. Those aren't things I associate with Go. That said, I think this particular implementation's client is also written in Javascript, so I don't know that it will have a particular advantage there. But an open source implementation at least won't block alternative clients (e.g. https://www.kickstarter.com/projects/2137936555/taut-the-fas...), in a way that the Slack Terms of Service would seem to do.

The desktop clients are built on Electron [1]. I've been using it for about 3 weeks and there's been a few times where performance issues were noticeable but not bad enough for me to start digging in the code to find the issue. Overall I think it's a well put together app, but people who hate anything built on Electron will probably have bad things to say about it.

[1] https://github.com/mattermost/desktop

Slack's performance also depends a ton on how much your teams are using it, and how many teams you're on.

"reactjis" are one of the worst offenders -- I've had a channel open with some people messing around with them, and it'll be casually consuming 30% of cpu and 3 gigs of ram on my top-of-the-line macbook.

I would have long since deleted an IRC client that was such a poor performer, but unfortunately there's not any good/complete 3rd party slack clients.

Lots of poorly written webapps give Electron a bad name. Slack, unfortunately does have it's performance issues, but there are other popular chat apps written in Electron that are very performant, even on large teams.

You can just use an IRC client and the IRC bridge...

Hey - i made that kickstarter! Unfortunately it doesn't seem there was much interest so it went nowhere. I still think it's regrettable how resource-hungry the official client is, but doesn't seem that too many people share that opinion.

I audited Mattermost 1.x through 3.x a while back. Had some issues but everything was fixed smoothly and from 3.x it is actually really good. Slides from my talk on it here: https://speakerdeck.com/addelindh/hacking-mattermost-an-asse...

Thanks @addelindh! Really love your work. You helped us a ton, highly appreciated!

Thanks yourself, you and your team is a pleasure to work with.

I believe you host the product yourself -- so if you are going to have to deploy it, modify it, etc., you'll want to know what stack it's built on.

Yep.

I still hesitate to even consider deploying rails or node.js applications.

It's just too many things to learn and keep up to date with.

Of course if you already develop in rails or node.js or run some other services already that might not be an issue.

But to me - unless it is apt-gettable - a standalone binary (typical go), war or jar file has a huge advantage.

Mattermost actually is apt-gettable if you install it alongside GitLab CE with GitLab's omnibus package¹.

1: https://about.gitlab.com/downloads/

And it's written in golang, not rails or node.js

That's just a throw away sentence in the article to emphasize its open source nature. All the main points of the article are bolded subheadings.

I disagree. The language the app is written in speaks a lot about the culture of the software. If it was Java, I would expect something very different than if it was written in PHP.

If you think they're all the same, then maybe you need to get some more real world experience.

Slack is written in PHP, I guess people care more when they have to host it themselves. I wonder whether the stories of Slack's codebase being a hot mess are the real reason it's still not available to self-host.

Its realtime server is Java though.

Implementation details can be important features of open source products.

This is a really good question, I'll try to answer from different perspectives (and hopefully others can chime in with what I missed).

1) For End Users and Developers, Golang can mean fewer bugs

Golang is a "type-safe" language, meaning that all the programming code is written to be explicit and consistent, versus "untyped" languages like PHP or JavaScript, where programming is more implicit with many different ways to say the same thing.

It's easier for software tools to automatically analyze type-safe languages and find bugs. And those languages are also easier to read and review--since there's only one way to express something--and that helps too.

2) For System Administrators, Golang means easier install and management

Golang is a "complied language", meaning that after the programming is done, a single file is produced. So a System Administrator just installs one file and Mattermost works. To upgrade, they just change that one file. Easy!

Languages like PHP and JavaScript are "interpreted languages", meaning that a System Administrator would need to download lots and lots of files, install one or more programs to interpret those files, and hope that nothing breaks. When it's time to upgrade, she's got to make sure all the files and interpreters on her server are exactly the right version. It can be a nightmare to maintain.

3) So what's the trade-off?

The trade-off with Golang is that it takes more work to learn than PHP and JavaScript. The programming language isn't as forgiving to new programmers. You need to say things exactly the right way or the program won't even start, whereas PHP and JavaScript if you say things "kind of right" and things will "kind of work".

Does this help?

Me and my team are planning to move to mattermost, but their Slack import just isn't 100% yet, so we've been holding off on it. For example right now it doesn't import the chat logs of any bot account.

https://mattermost.atlassian.net/browse/PLT-48

While we're not in the biggest rush to move to Mattermost, it does seem to offer more than Slack right now.

Do you really need the backlog of bot output for a migration?

For me the biggest feature Mattermost has over Slack is that you can host Mattermost yourself.

If you use GitLab or want to consider using GitLab, its omnibus package comes with Mattermost included. A really nice combination.

I'm concerned about the mobile-friendly options for MM. Slack has pretty good iOS apps (and others, but I've never tried them). Does MM have quality mobile apps?

The android app is basically a webview and not very useful.

On my android phone, the mattermost app does something weird to my keyboard - it replaces the right-most 1/3 of the spacebar with a key with a '->|' symbol on it that closes the keyboard when pressed. I've never seen any other app do this, and it's (of course) really annoying because it means when I try to hit the spacebar the keyboard closes instead.

That's just one of a couple issues I have with the app (push notifications not really working, the chat taking 15 seconds to reload everything upon switching back to the app, etc) that have basically made it useless to me.

@i_are_smart, ouch, that keyboard thing sounds really weird. Would you be open to joining our public test channel to share a screenshot? https://pre-release.mattermost.com/core/channels/public-test...

Would highly appreciate your feedback on notifications and load times as well.

@it33, apologies for the slow reply here, I basically never check my messages. I'd be happy to do that once I get some time at work, or afterwards.

For the sake of replication, my keyboard is "Hackers Keyboard", running on Android 5.1.1 (specifically OxygenOS 2.2.1) on a OnePlus 2.

still testing out mattermost, but ios app seems great so far. very similar experience between web/desktop/ios apps.

I use the Mattermost iOS app daily, and it works well on a good connection.

The iOS app is a feature rich experience similar to web and desktop apps, so when I travel and am on a slow connection it doesn't perform as well.

Here are our current mobile performance benchmarks so you can decide for yourself: https://docs.mattermost.com/deployment/push.html#mobile-perf...

There's work underway to get the mobile apps working better on slow connections.

You said it's feature rich, therefore has lower performance over slow connections.

This is not true. Mobile app performance over slow connections is determined by app/server architecture, payload efficiency, caching strategy, etc.

There are many examples of feature rich apps performing well on slow connections, and many simple apps performing poorly over a normal connection.

>What if Slack had threaded messaging?

Then it would be a recipe for endless sub-discussions diverting from the topic, and disaster...

We had this in flowdock before we switched to slack, on a team of probably 700 or so at the time. It made it far easier to filter out chatter when dealing with an urgent issue (no need to change rooms or remind people to be quiet for a while), and I haven't really noticed it encouraging pointless chatter. I seriously miss this feature in slack.

What was wrong with flowdock?

I'm not sure. I think with the benefit of hindsight, I liked it better on balance, but I wasn't involved in the decision. It was definitely a terrible resource hog, even more so than Slack, and the iOS client wasn't quite as polished IIRC.

that doesn't occur without threaded messaging? threaded messaging makes it easier to ignore those sub-discussions...

>that doesn't occur without threaded messaging?

The difference I mentioned wasn't one of occurring or not, but one of degree of occurrence.

>hreaded messaging makes it easier to ignore those sub-discussions...

Not as much as it makes it easier to start and digress into endless sub-discussions.

Wouldn't that it is self hosted defeat the point of single sign in across teams? I mean I can see if the hosted instance has many teams you subscribe to, but in the Slack world I already manage sign ons across different company/project/team domains, so the advertised advantage here doesn't seem to apply?

if you need to connect to a team that is not on your hosted server, you can still do so via the web or the desktop client either in a separate tab in the browser, or in a separate tab in the desktop client.

I thought they were going to compare it to matrix/vector.im.

Mattermost is open-source alternative for slack, matrix is infrastructure for IoT using websockets.

I love that there are some other tools coming up to challenge Slack -- hopefully this will put pressure on Slack to up their game. Slack has been... very slow to fix bugs / add new features. They're starting to feel like Microsoft... like they don't move fast, aren't trying to improve anything, and don't care about customer feedback.

I love Slack, but some of my gripes with Slack:

* I can't spell check from a Post in Slack. Posts would be great for taking notes, but they lack order. It doesn't have to be full-on Confluence / Wiki style, but being able to add some classifications to Posts would be helpful. And in-line spell check... Every other tool everywhere supports this, but not Slack Posts.

* Posts and Chats use different formatting markup. Why not just use Markup for both?

* Channel names being forced into lower case and very short character counts... Why can't I use a longer name for a channel and just wrap the name in the display? Why does it have to be all lower-case without spaces? Why isn't there a folder structure to the channels to keep all my clients / focuses grouped?

* Slack forces me to put Bots and People in the same channel and it gives them both equal visibility. I should be able to add a #bot or something and have those messages be something I can search for, or something that shows up in the sidebar, but not something that talks over people. Their current work-around is to use two channels, one for people and one for bots... but making sure your team is added to every channel... it sucks. FlowDoc does a better job of this, and if it was built the way they do it I wouldn't have to search multiple channels to find the info I could just search one place.

* File storage in Slack is a nightmare. Try finding a file again a month later. If I upload the same file name why can't it just version the file like Box or RedPen does? Nope, it just uploads the same file again... tagging a user or # doesn't actually tag the the file the way you want it to... so I can't even really search by those things.

* When you click "Open" a file you have to log in again? That's so busted. I can download a file, but if I'm on a phone and I click "preview" it prompts me to sign in -- extra tedious for users who enable 2FA. Security doesn't have to be that tedious, just let me open files without a login -- if I can download them but I can't open them directly it's not security, it's just an annoyance.

* The default integrations in Slack kind of suck. I know that's not 100% Slack... but why can't I get updates when someone makes a change to a Google Drive file, or creates files in a Google Drive / Box / Dropbox folder?

* Hashtags are busted. Why don't # work like they do in Twitter where I can search for them after? Having # be rooms isn't good. Nobody gets that to start... I'd so much rather be able to tag a conversation (oh and have a threaded conversation) around a # and then have that # show up in the channel column...

* User management is lame. As the administrator, why can't I go into a user's profile and add them to multiple rooms at once? Or why can't I add a bunch of people to multiple rooms at once? Also why can't I lock down user names -- I don't want my users to change their user names to political statements or stupid handles, if I set it First.Last I want it to stay that way. The last thing I want is for one of my developers to rename himself @ZombieGoatMaster4DonaldTrump and have that be something that is shown to a client.

* Mobile isn't consistent with Desktop. Why don't the channel lists in Slack sort the same way on Desktop and Mobile? Why does the + sign mean join a channel on Mobile, but create a channel on desktop? It's so inconsistent it's like they don't even bother having their different product managers talk or work with the same UX team.

Anyway really glad people are producing alternatives to Slack. Going to check this out, thanks!

> Slack has been... very slow to fix bugs / add new features. They're starting to feel like Microsoft... like they don't move fast, aren't trying to improve anything, and don't care about customer feedback.

we're looking for a self-hosted alternative. tried hipchat, and it was worse than slack in the areas you mentioned. mattermost has been a ray of sunshine, so far, relatively speaking. still testing it out, but they are super responsive to their users and community, between github, their forum, their pre-release server (https://pre-release.mattermost.com/core ), etc... they also seem to be aggressively adding new features and fixing bugs.

I realize I probably sound like a shill, but it just has been that much of a positive experience so far.

I hijacked your post, sorry. Will give Mattermost a look!

EDIT:

Your site doesn't share correctly.

I went to share this on LinkedIn:

* What Slack might learn from its Open Source alternative | Mattermost || https://www.mattermost.org/what-slack-might-learn-from-its-o...

And I see this:

http://imgur.com/a/evVZF

You're using WordPress so you can check out:

* Yoast SEO — WordPress Plugins || https://wordpress.org/plugins/wordpress-seo/

This will add the right meta tags to make the thumbnail and image and description show up through shares with minimal effort on the part of the content creator.

Yeesh, wow, thanks @dbg31415! Have mentioned our Wordpress person to help fix this, appreciated!

How about Discord?

Discord is great, but its primary purpose is for gamers, but can of course be used for similar purposes as Slack.

The problem with Slack & Discord that Mattermost addresses is the ability to host it yourself, which means that you're in control of the data. Not everyone needs that, but there are companies and especially government-owned institutions/companies that do require that.

No integration support is the boon of discord for software teams.

comment threading is kind of amazing (in mattermost).

Thanks!

Yeah, we love using threads to keep discussion on track.

It took a long time to design. We weren't able to find anything like it, and had to design through a lot of corner cases.

We're still working on making it more discoverable. Once you use it, it's hard to go back, but it takes a while for people to discover the "Reply" button on messages.

Reply on news.ycombinator.com