3 ms·
This is an extremely unpopular take in today’s world where 99% of what you use is expected to be “free” and subsidized by those with interests thoroughly misali
by strogonoff 2y ago
This is an extremely unpopular take in today’s world where 99% of what you use is expected to be “free” and subsidized by those with interests thoroughly misaligned with yours.
Nevertheless, it is my strong opinion that offering paid service is the only way to a healthy ecosystem, healthy competition, products that actually benefit the user, and the market that works as the market is supposed to (and that Big Social innovating[0] the subversion of this model is a tradegy with net negative effects on society).
Not offering paid[1] service to customers means ultimately either 1) dying or 2) making someone else your paying customer (e.g., the advertisers being offered the users as the product).[2] That “someone else”, by being the source of revenue, will dictate how the service evolves, and their interests will not be aligned with the users’ except by sheer accident.
In case of services like Bluesky, I doubt API access can be excluded from the equation, since it costs money, and when inevitably abused a lot of it, sadly.
Federation overhead could in theory be subsidized by paying customers, or there could be smart controls such as letting certain servers federate for free if it benefits paying users while charging others an honest fee[3].
Finally, let’s not forget the good old “charge for the app”. Sure, being open means someone can build another, cheaper app, and someone will use it, but believe it or not—that’s normal competition; make your app better, respect your users, turn profit.
Bluesky’s accepting large investment is a signal of not going in this direction, and is therefore worrying.
[0] It could be said that some companies (e.g., Adobe and Microsoft) did something like that earlier, subverting the market via offering a product for free by not seriously chasing pirates. When someone’s choice is between two paid products—instead of free awesome product everyone pirates vs. some mediocre paid product—the competition works much better. However, their core offers were still paid, and crucially those were not social media services that prey on our attention and anxiety.
[1] Before someone mentions the disadvantaged: offering subsidized options to those who need it is not mutually exclusive with a straightforward commercial model.
[2] There’s also 3) becoming another Wikipedia, which happens every so often… Let’s be realistic.
[3] A lot of courage would be required, as I can already hear the “walled garden” outcries.
- ianopolous 2y agoVery well said. This is the approach we're taking with peergos, paid accounts unless you self host. I wrote some of our thoughts on this here: https://peergos.org/posts/the-social-dilemma https://peergos.org/posts/the-social-dilemma
- strogonoff 2y ago1. I am not sure it is possible to compete with a free product. By definition it will have more users. That’s the tragedy, competition is not possible. 2. My approach is not “one more product will be the answer”, more like “fixing the incentives and making the market work will enable many products”.
- ianopolous 2y ago1. It depends on the time scale. The question is if you can outlive their VC funding. 2. How do you fix the incentives short of fixing capitalism to remove software VCs?
- strogonoff 2y ago1. They are supported by paying customers (the advertisers), not just VC money. 2. You don’t need to get rid of VCs, it’s orthogonal. Just make interoperation/complete API coverage required by law to be available to any user if you have more users than certain amount. It would neuter the lock-in when anyone can make a third-party GUI that has no ads and interoperates with any other social platform (meaning no “all my friends are there” effect).
- ianopolous 2y agoYes, legally requiring open APIs and interop beyond X users is the best way forward. There is some tension with allowing product changes and deprecating old APIs though. Maybe something like mandate versioned APIs, and keep old versions around for at least 2 years after newer version is released.