Show a server list at first launch

[niwla23]

I recently saw this in another client:

So they have a server chooser built into the app. This is pretty cool to promote federation.
It would be cool if this could be integrated into Element too, since matrix.org is really dissapointing when it comes to speed.

So you would have a list of trusted servers and then the client will ping them and sort them by ping

[nkosi23]

I feel this would be confusing and not user friendly, especially for newcomers. To really promote federation, there should only be 3 choices:

1. the matrix.org server,
2. the server operated by EMS (so that they are rewarded for their work),
3. and the option to use a custom server

Homeserver owners should really advertise their value-added out-of-band (for example on hobbyists forums used by the communities they target + using a marketing website created by the homeserver owner). This is facilitated by #2662. Among many other things, this will allow homeserver owners to have more real estate to make their case. This is similar to what is done with free email providers (open-source email clients do not present a list of dozens or hundreds of free servers).

My thinking is that ultimately no one would truly benefit from having a list of 10, 50 or 100 servers as a home page. Such a splash screen will make Matrix and Element look like something like IRC (a tool for geek insiders). I think the few people who would truly be interested and helped by this list would be better served by having a wiki website referencing these homeservers, especially since copy-and-pasting a server config will not present any difficulty for this kind of audience.

[ManDay]

I'm compelled to comment because I'm trying to assess FTUE at this moment. I don't think the Matrix ecosystem will ever work by advertising homeservers over protocol. Because Matrix is federated, people approach it with the goal to join Matrix, first. They do not come to Matrix with the intention of joining a particular HS in mind. They don't care about servers or infrastructure. These are, understadably, implemention details to them.

People want to join Matrix, they install the client, and with @nkosi23's suggestion a 95% of them do inevitably register to matrix.org, as they currently do! While with @niwla23's suggestion they are faced with an implementation detail that will possibly irritate them.

[nkosi23]

Matrix federation is similar to the email network, it is useful to keep this reference in mind to understand the dynamics at play and how to most effectively solve this challenge.

Who advertises email to end-users? It isn't postfix or sendmail. It is Gmail, Office 365, Yahoo Mail etc... So why expect Matrix or Element to do so, especially since this gets out of control essentially immediately, leaving everyone dissatisfied.

Email providers are the ones who really market the email network through their hosting offerings (disk space, plugins, addons - for example calendar - , ecosystem added on top - do you remember that many people used to create hotmail accounts mainly because they enabled them to use MSN Messenger - etc...). Similar ecosystems will emerge for Matrix with every HS offering different disk space, bots, communities targetted / brand positionning, proprietary bridges to proprietary ecosystems etc... Such a marketing must be done out-of-band, because running a HS is essentially running a business.

Matrix is not like running a TOR exit node, or running a seedbox on the bittorent network. When you run an HS on Matrix, you are not and cannot be entitled to be discovered, because you aren't necessarily useful or helpful to begin with. You have to make your case and market your business and value proposition, just like you would for any other business. Expecting something different is not reasonable and would not work (cf the email network and the already overwhelming list in the OP).

This is the only way a federated ecosystem can be healthy. A client's role isn't to advertise providers (except if they have a commercial relationship with them, like Element and EMS Cloud for example), every provider must create value-added and do its own advertising out-of-band.

Creating value added and making a case is difficult, but this is yet another reason why it is not realistic or reasonable to expect a client to do the marketing of 100s or 1000s of providers.

The matrix.org server was/is necessary to bootstrap the network, but very rapidly we'll start seeing several HS providing much more value added than matrix.org.

I wouldn't worry about seeing the matrix.org server being the only server used in practice. As adoption picks up, matrix.org will be forced to resource-limit and rate-limit increasingly aggressively, and market matrix.org more like a demo server while actively inviting people to find a proper HS. We are still in the early days of this journey, let's give it another 5 to 12 years before we reach this point, but right now it is in the best interest of everyone to have a go-to matrix server available, until a dozen or so key HS emerge, and the general public becomes as familiar with the matrix ecosystem as they are with email.

I see no risk of matrix.org becoming the Gmail equivalent of the Matrix ecosystem. They do not have the funding required to begin with, but this also clearly isn't what they want. I am therefore confident things will go in the direction I described. And the status quo, does not prevent homeservers from successfully marketing themselves out-of-band, on the opposite this is a fine balance.

What could be useful and viable is a website / online directory referencing homeservers. This would be a better approach to support the ecosystem than a splash screen. I'm sure the Matrix.org website would be more than happy to reference quite a website, but I personally do not think that Element should link to this website for reasons already explained in my two messages.

Hopefully this clarifies my thinking.

[niwla23]

You compare matrix with email, which is a good comparsion IMO. And many mail clients HAVE a selection. You always have the top providers there: Gmail, Yahoo Mail, etc and the option to use a custom server. I think this never confused anyone, nearly everyone knows that there is no central "email" server and you need to choose a provider (even though it is often gmail, just like matrix.org).

[niwla23]

And I think that matrix.org is bad for newcomers, because it is really really slow. People will say "Matrix is slow" and never come back, when they actually mean: "matrix.org is a slow homeserver"

[nkosi23]

To add to the list of pros and cons, today Element just got suspended from the Google Play Store, and it seems that the reason for this is that someone flagged inappropriate content and Google decided to consider that Element is a content platform rather than a client allowing users to access a federated network. This is like the law: you can frame yourself and your actions under any light you want, but at the end of the day the judge will frame you in light of the law of his land, and he will see whatever his interests require him to see. This is the world we live in.

In this context, linking to servers having each their own content moderation policy will further multiply the risk that this kind of unfortunate event happens. Now one will say: "this is not a problem, let's only list servers adhering to a content moderation policy forbidding hate speech and such and such". Well, congratulations, you have now turned Matrix into a centralized ecosystem similar to Mastodon where once far-right activists migrated to this supposedly decentralized technology, Mastodon as a network was forced to implement virtual content bans and pressured client applications into implementing blacklists.

This kind of initiative introduces plenty of questions like: who will maintain the blacklist, what are the rules to be listed, what about free speech etc.... This introduces a single point of failure and weakens the DNA of the Matrix ecosystem. Let's not forget that the goal of Matrix is to be far more than a messaging infrastructure, the goal is to provide infrastructure for arbitrary data, so we shouldn't err too much on the side of trying to make a social network, especially at this bootstrapping stage.

To put it differently, it is not reasonable to take the risk of seeing Matrix fail because of content moderation controversies, considering Matrix grand plan (in addition to my previous arguments). To me, the best thing to do is to educate users and get them into the habit of looking for a "Matrix Provider" out-of-band, just like they look for an email provider out-of-band.

If we want Matrix to thrive in the long-term, at this point the discussion should actually be about removing even the two links to matrix.org and EMS, and implementing deep-links as per this feature request to make server discovery entirely out-of-band.

Regarding the fact that many email clients have options to select a predefined email provider, let's remember that none of them allows users you to create an account from within the email client. The user is supposed to have discovered the service out-of-band and already have an account on it. Predefined providers are listed in email clients not to give them free advertising but to attract their large user base and make it easy for these users to use the client. This is a clearly different motive and situation.

But even putting this aside, we are no longer in the early 2000s where people were unaware (and , too often, naive) about tech business models. Nowadays people are suspicious and the gatekeepers like Google are under pressure, so they are taking an increasingly aggressive stance on social media moderation, and anything looking remotely like a social network will face intense scrutiny. This is a fact that must not be ignored in this discussion.

For the reasons I've explained in this thread, as far as I am concerned, I feel this server list is false good idea. There is too much downside, and I feel it would lead Matrix in the wrong direction. We do not need another Fediverse, especially not in these times of free speech / moderation controversy. Implementing this feature would be like buying trouble sold at discount.