Why not just enable #mastodon instance moderators to mark some of their existing lists as "starter packs" and offer them to newly registered users for automatic follow + list import at once? (Without need to save and import CSV manually...)
not exactly algorithm, but as you mentioned - curated lists, named, with icons, which beginners can automatically join. But these lists would be instance-local and available only on instances, where moderators choose to offer them (and I hope they would choose wisely)
Very often, the feedback from new users is "but all I can see is empty web page, it is boring". The only place from which they can choose is local timeline (boring) federated "other servers" timeline (very confusing, for newcomers) and then maybe the "Explore" option.
But I think, that having some starter packs, like "Nature photographers", "cycling" or "hiking" would be safe. People with specific interested would be able to relatively easily join some active groups. It could respect the toggle "make this account discoverable for others" - such accounts would be excluded from the "exported" lists.
Besides new users, all instance users would see curated lists besides their own lists (somehow, or maybe in the Explore section) and would be able to mass join curated lists (and perhaps also mass-leave, and also convert to personal list, so they can separated imported follows from their own follows).
Yes, there is risk of creating echo-chambers, but the diversity of such lists among instances would make it safe.
If a new user signs up nowadays, they will see a list of suggested accounts to follow which is automatically generated based on which accounts are most followed on that server.
Also, the admin of the server can customise this list of suggestions to add or remove particular accounts.
if I can customize this list in latest versions, I am probably not aware about it!
But I create many lists on-fly, just sorting the newly discored acounts of lists. Some of these list are pretty hamless and would provide good "starter packes", which can be mass-followed by anoyone on instance (but also hidden from main timeline, perhaps). I mean groups like "Nature photographers", "City photographers", "Generated art", "Travellers", "Hikers", and so on - there are many accounts, which really provide just one type of content and can be easily categorized.
As instance moderator, I am getting lot of feedback, from both new and more experienced users. New users find it hard to find any interesting content to follow: we don't want algorithm, so creating named specialized lists manually may be the other way?
Also users requests chance to tune the timeline. There is already "hide list" feature, I use it to mute too much active accounts. Other users request feature to hide boosts and to see only original content (perhaps could be also property of certain lists? just like hiding, but hiding only boosts?)
Ok, I should give this feature a try. But I can't really think of specific !10 or 20 accounts I would recommend to all new users, especially as we are general purpose instance, not single major topic instance.
Instead, I would just like to make SOME of my Lists (which I use heavily to categorize the content I discover) shareable as a whole (which they in fact are, but using CSV, which is easy for me as programmer and PC user, but quite cumbersome for typical smartphone user...)
So instead of recommended accounts (10 or 20 is not enough), we would have 10 or 20 recommended "starter packs", lists of many accounts to follow at once. And also... existing users should be able to see these share or exported lists in their "Lists" or Explore section (simply somewhere in between, maybe even as subsection of Live channels... finding right place for new feature in GUI is tricky).
And instances could differentiate one from another by these starter lists for newbies being VERY different...
Definitely, these lists would be man made (moderator already have some power anyway), so it is not algorithm at all. It is less algorithm based than Explore. And it would be completely voluntary feature, not enforced on anyone.
If you have a website or whatever, there is nothing to stop you offering CSV files to your users for them to import. You can make an announcement about it using your admin announcement system.
That happened when Mastodon exploded in 2022, for example journalist were exchanging CSV files of other journalists to follow.
@FediTips yes, exactly. We had this in czech Mastodon community too.
All I ask for is to integrate this feature directly inside web GUI of Mastodon, perhaps somehow recycling or reusing the lists interface. So I can share the lists only by clicking inside GUI and not exporting and importing from external files.
Which is also very good and useful, but probably not for beginners, especially younger users. Kind of CSV import wizard dialogue, maybe :-) With lot of icons, less words.
I am asking for the existing CSV import/export feature to be integrated with the GUI (without need for CSV files to actually exists, but who knows).
Instead of standalone recommended accounts, instance admin or moderator would recommend entire lists of interested accounts to follow. Like algorithm did on Twitter. But it would not be algorithm, but moderator's choice.
Because new users know nothing about recommended accounts. The choose randomly and it sometimes acts as echo chamber.
Moderator can recommend accounts with very little followers, which is actually good. Moderator is person with enough time to spend lot of time exploring Fediverse.
Because it is anyway what advanced users actually do, but for average users it is too complicated.
I just suggested using Lists infrastructure for this, because it is already there, there are some dialogues and so. So it would be easier to recycle user lists, instead of implemented some moderator interface for recommendations for scratch.
Also, it would be nice to be able to see dynamically generated lists, by the way. Eg. "all the people I fav most, all the people I boost most" and so. So I can return to these people and add them to my lists :-)
I just don't want all the time we waste on social networks to be wasted. Browsing the fedi content is real work and I would like to share results of this work efficiently. Boosting too much is not the right way to go.
Sometimes, waiting for the natural process, eg. the good content being boosted enough times, takes years... and for new users, it may be hard to initiate the process at all...
I think there has historically been danger of CSV files being used for spamming?
There was one major Fedi directory that used to offer CSV files years ago, sorted by topic, but they had to stop that because they were being misused by spammers to target particular accounts.
this is what I answer to them, especially using tags... but still, it takes some time for new users to follow enough accounts to build timeline in which at least something really happens.
I can share my lists using CSV export, but importing CSV lists manually is extremely unfriendly. So I am not requesting feature which is not there - CSV exports and imports of the lists are already there - I am just asking, if the existing feature can't be done more user friendly, with the lists maybe being even shareable objects in the timeline? and explorable before following?
I think the issue is with users coming from an algorithm based platform. From new it took me about a week to fill up my Social timeline, up until that point I was lurking Local and General.
I still go into General to see if there is anything interesting out there.
@FediTips maybe, but I am not part of the developer community, so I am broadcasting the idea, so it may reach someone from the community.
Mastodon source code is incredibly complex and deeply structured; I know it, because I did some well documented little hacks, like increasing maximum status length or poll option count or changing Publish! button back to Toot! So the changes are not easy from someone who is new to that.
Importing existing lists marked as "starter packs" by instance moderators as list of followed accounts should be technically very easy; it is just the complexity of the code, which requires this to be done by experienced Mastodon contributors. Most importantly, there would need to be consensus in Mastodon community, that it is right thing to do...