H
So if we really believe in a free market, we should believe that markets being made cannot be free, for the user, the producer or the publisher. They can verily be systems that are predatory and exploitative of one or more parts.
We cannot exploit the user, the creator or the publisher. This cannot be overstated. In the west it's gotten too normal to centralise ownership and to exploit people. This is something we have to be mindful of.
If we then come up with some great idea for say a client side algorithm, or some alternative ad networking technology, you can then wrap that up in a plan for an organization and apply for funding.
From that you can start figuring out the various ways to make a business off the resulting systems.
But they have to be open, give power to the user and allow for competition. This is why #OpenSource is so important, because it levels the playing field.
We can fall back to Silicon Valley, we can fall back to Wall Street, we can fall back to London City, but the great thing about the fediverse is that people can ignore all that, filter it away, etc.
People don't want to go back to #PredatoryMonetization - and nor should they.
This means that we have to work within those parameters, and there is the chance to actually innovate and quite possibly disrupt markets - in a good, ethical and moral way.
I think the one thing you have to be aware of in regards to denizens of the #fediverse is that they are verily fatigued by large commercial platforms because of the #ads, the predatory #algorithms, #darkdesignpatterns, #tracking all stemming from the need to #monetize.
Each of those points are important because something's shouldn't be monetized, like #databrokering.
That's why I am suggesting client side algorithmic discovery, or a private and safe alt.
But then my idea hit a wall, because I suddenly realized how big the fediverse could get and what burden it might put on platforms.
How about this though: spin off the discoverability database to itself and maintain it as an #OpenData project.
Anyone who wants to participate could mark their account or content as discoverable and be included.
There could be a CoC to prevent abuse and guarantees that creators wouldn't be excluded by virtue of licensing.
The big iffy issue about algorithms is how it gives vendors too much power. Basically a vendor can "play favorites" and prevent certain content creators from being discovered, or push their favorites.
Consider how vendors like YouTube and Instagram started to push the creators they wanted you to see - despite what the user might want to see.
In essence, my biggest reason for client side algorithms is to take that power out of the vendors hands.
Or is there a way to compartmentalize data and interests into a smaller subset? Even binary distributions of content lists.
I think we need some #DataScience people onboard.
How does one ethically collect data or split up the job so that the user can maintain privacy?
I think "the other side" has valid points, and that's why I'm thinking of client-side discovery algorithms - but it's not an easy problem.
If the client is the one who is doing the sorting and selecting, it means the server will have to distribute MASSIVE amounts of data all the time to clients so that user preferences can be widdled down on the users side.
But is there a middle way? Can the task be split up so that some if not all privacy is maintained?
Unfortunately I don't know if anyone else is talking about this, mostly because one side is against #DiscoveryAlgorithms since they feel typical sorting functions should be enough, and the other side thinks that letting clients handle it would mean too much complexity and bandwidth usage.
There are also those who hate any form of algorithms spreading content at all, and those who want their algorithm to be secret sauce.
The #fediverse needs to come to terms with one thing: on content platforms, chronological ordering is probably good for your subscription feeds, but for the #discovery feed the average user actually WANTS an #algorithm.
Instead of kicking your feet in anger, release your anxiety and think for one moment.
Wouldn't it be better to formalize #discovery algorithms as #opensource? To create programmable interfaces server side so that the user can define their algorithm client side?