I think Reddit’s CEO is making a fool out of himself by how he’s managing this situation. I think however that the solution is very simple and straightforward.
Let’s start: I can understand that Reddit has costs to operate the platform. I also get that they don’t want big companies to abuse the API to train ML models and profit of it. Fair game!
But why not offer a generous free tier for regular users? Say, every user gets 500 free API calls per day. Regular users stay within the free tier, while big companies can’t do anything meaningful with only 500 calls per day (so they end up paying money).
Seems pretty straightforward to me. Everyone happy! Many other companies offer generous free-tiers for exactly this reason. Am I missing something?
My impression is they’re being disingenuous, for the reasons you say. They could easily support 3rd party apps but ban large-scale data mining. Saying “supporting these apps costs us money, so we need to charge” is a manipulative half-truth. Like Selig said, they’ve priced it not just at covering their costs but making a healthy profit.
It’s not clear why they don’t just serve ads in the API and require them to be displayed, or implement profit-sharing with 3rd party devs (as in, they pay reddit a portion of their income from ads/subscriptions). The only clear reasons would be for control and to pump up numbers for the IPO.
Reddit could let users access the API like this easily. They could stream ads along with the comments coming from the API. They could let individual users pay a subscription fee for their own api access. They could develop an advertising platform for 3P apps to show reddit ads.
They could even have said: look, we’re going to kill off 3P apps because we have another idea now, thanks but you are no longer required. At least that would have been a genuine approach.
Spez evidently has an idea about what he really wants and isn’t sharing it yet. I’m sure it will be clear after the IPO.
They looked at the numbers, concluded that 3d party apps were a fringe phenomenon that could threaten their control over the platform, and just killed them.
There are many possible revenue models that include 3d party apps and a more open API, Reddit just isn’t interested. They see Twitter as a shining example for some reason.
I still believe that the ML companies “argument” is just a giant smokescreen. Reason is simple: ML companies can, and probably always have, just scrape the website. Why build an integration for every API under the sun if you can just build a web crawler once and be done? There are even existing, free implementations available so that’s an absolute no-brainer.
It’s about killing independent clients, nothing else.
Actually when I think about it you are absolutely right. The ML argument is complete bullshit. I mean to train a ML algorithm an API is nice but scraping should do just as fine. I don’t know how complicated the Reddit API is but you essentially need just GET so I guess not that much. How much time would a development team need to switch the implementation from API to scrape? A week? We’re in corporate world so let’s say a month with all the corporate bs around. That’s still nothing
I think Reddit’s CEO is making a fool out of himself by how he’s managing this situation. I think however that the solution is very simple and straightforward.
Let’s start: I can understand that Reddit has costs to operate the platform. I also get that they don’t want big companies to abuse the API to train ML models and profit of it. Fair game!
But why not offer a generous free tier for regular users? Say, every user gets 500 free API calls per day. Regular users stay within the free tier, while big companies can’t do anything meaningful with only 500 calls per day (so they end up paying money).
Seems pretty straightforward to me. Everyone happy! Many other companies offer generous free-tiers for exactly this reason. Am I missing something?
My impression is they’re being disingenuous, for the reasons you say. They could easily support 3rd party apps but ban large-scale data mining. Saying “supporting these apps costs us money, so we need to charge” is a manipulative half-truth. Like Selig said, they’ve priced it not just at covering their costs but making a healthy profit.
They priced it to destroy third party apps. There is no other reason for doing it the way they did:
Yeah. They want to kill the third party apps so everyone has to use the ad-supported Reddit app.
It’s not clear why they don’t just serve ads in the API and require them to be displayed, or implement profit-sharing with 3rd party devs (as in, they pay reddit a portion of their income from ads/subscriptions). The only clear reasons would be for control and to pump up numbers for the IPO.
They’re likely more interested in control than revenue.
Reddit could let users access the API like this easily. They could stream ads along with the comments coming from the API. They could let individual users pay a subscription fee for their own api access. They could develop an advertising platform for 3P apps to show reddit ads.
They could even have said: look, we’re going to kill off 3P apps because we have another idea now, thanks but you are no longer required. At least that would have been a genuine approach.
Spez evidently has an idea about what he really wants and isn’t sharing it yet. I’m sure it will be clear after the IPO.
They looked at the numbers, concluded that 3d party apps were a fringe phenomenon that could threaten their control over the platform, and just killed them.
There are many possible revenue models that include 3d party apps and a more open API, Reddit just isn’t interested. They see Twitter as a shining example for some reason.
I still believe that the ML companies “argument” is just a giant smokescreen. Reason is simple: ML companies can, and probably always have, just scrape the website. Why build an integration for every API under the sun if you can just build a web crawler once and be done? There are even existing, free implementations available so that’s an absolute no-brainer.
It’s about killing independent clients, nothing else.
Actually when I think about it you are absolutely right. The ML argument is complete bullshit. I mean to train a ML algorithm an API is nice but scraping should do just as fine. I don’t know how complicated the Reddit API is but you essentially need just GET so I guess not that much. How much time would a development team need to switch the implementation from API to scrape? A week? We’re in corporate world so let’s say a month with all the corporate bs around. That’s still nothing
Maybe he’s so incompetent that he honestly doesn’t know that machine learning companies don’t need api access to do what they do