As guerras de API estão aqui

The API Wars Are Here

Find out why social media giants like X and Reddit are changing their API policies and what it means for the industry. Discover the impact on small businesses and get valuable advice for startups navigating this changing landscape.

Imagem em destaque

To quote one of the funniest television shows of the last decade: “This is the darkest timeline.” Ever since social networks began to exist, services like MySpace or Friendster (which should tell you how long ago we've been talking) have had APIs for third parties to play with.

Permissive APIs enable very creative plugins and adjacent applications created by the tech-savvy community. It's the lifeblood of your service when your team doesn't provide continuous delivery. Power users create something that suits their preferences and share it with the wider community, other users stick to it, and some even build on top of it to provide a more refined experience.



If APIs are such a good thing and have been the industry standard since their early days, how come two of the biggest social media companies, X and Reddit, suddenly decided to do a 180° turn and close the gates – to the point of killing adjacent services and applications that were loved and widely used by the community?

Well, put on your fedora hats and grab your best magnifying glasses, because today we're going to explore the reasoning behind these decisions and the impact they could have on the industry at large. Let's dive into the big API wars of 2023.

Policy change

The migration of companies to a different policy is nothing new. Perhaps the most glaring example is Google . Initially, Google adopted open source principles for Android, but as it gained market dominance, it gradually moved away from open source ideals. Additional development and features have been added to a close code alternative. Without a dedicated team of community developers, the application will end up as abandonware

Google wasn't wrong here. When they launched Android, it was against the titans that iOS was up against at the time. With no market share , the most important step was to gain a foothold, no matter how small. The target at the time was a geekier, more tech-savvy crowd, in stark contrast to Apple's fantastic user experience and more welcoming experience.

But that was then, and now with over 70% market share , it's pretty clear that Android is in a very comfortable position. Android got to a point where it got so big that leaving everything open source would create its competition. So it was time to bring him into the house.

mobile operating systems

US market share in August 2023. Source: Stat Counter



We love openness, and from our perspective, open solutions make for better products . A larger community means we have more eyes to find bugs in the code and more people tweaking and playing with it, pushing its limits.

But at a certain point, a company has to start thinking about its competitors – not the developer community at large, but rather other Big Tech names ready to take a bite of the pie. It's the nature of the business. If you have a good product, others will try to learn from it, whether by looking at your code or hunting for developers. Open solutions are fantastic and empowering, but they can eventually become a liability for the business.

What is an API

API stands for Application Programming Interface. In simpler terms, the API serves as an intermediary that allows two applications to communicate with each other. For example, the Twitter API is a set of functions provided by Twitter (now X) that allows developers to access and interact with certain features of the platform.

The X API works as an information channel, allowing users, mainly developers, to retrieve X data or send requests for certain actions. It's like a waiter in a restaurant who takes your order and delivers your food. API X receives your request, delivers it to the Twitter server, and then brings the server's response back to you.

For example, when we use a third-party application like Hootsuite that needs to display tweets, it sends a request for tweets to API X. The API then retrieves the tweets from server X and provides them to the third-party application.

There are several uses of API X. One of the most common examples is automated tweets. Here we can use the API to schedule tweets in advance or automatically tweet content from an RSS feed. Another application is trend analysis. We can use the X API to collect data about what's popular or trending on X at any given time, analyze that data, and use the insights to make strategic decisions.

Social Media API Features

Resource To benefit
Detailed information about account activity User behavior, sentiment analysis
Easy to use User-friendly interface when accessing important features
Greater engagement Increase reach and improve performance on social media
Identify influencers Insights into user accounts to identify marketing opportunities

How AI started the API wars

OpenAI turned the world upside down with the public launch of ChatGPT. It wasn't the first language model, nor even the first implementation of GPT-3 (it was for 3.5, but that's the basis of 3), but it was the demo that blew people away with the sheer refinement of the generation. text.

ChatGPT became the app with the highest and fastest adoption rate in history, dethroning Instagram. This sounded alarm bells for companies who took note of what OpenAI was doing and decided to up their game. In the following months, Amazon, Meta, and other companies joined the race to build LLM solutions. Meanwhile, others asked questions.

So you have a fantastic model with billions of parameters, but where did the data used to train that model come from? According to a paper from Lambda Labs, the performance of a language model varies depending on the size of the model, the size of the dataset, and the amount of computation following an exponential relationship.

In other words, more data and more parameters equal better performance, reaching the point where these models are able to make educated guesses related to subjects they have not been trained on. Now consider that GPT-3's dataset and model size are about two orders of magnitude larger than those used for GPT-2 and you start to get an idea of ​​how much information has been fed into the model. Spoiler – it was 300 billion chips .

Companies soon realized that these tokens had to come from somewhere (e.g. from their services via APIs). . This might have been fine when OpenAI was a non-profit organization working on donations from other companies, but the minute they estimated revenue of $200 million from ChatGPT in 2023 and $1 billion in 2024, it was time to ask for a part of these profits.



It's always a question of money

As the saying goes, if you can't see how a company is making a product, you are probably the product. Social media is a complicated business; they live or die based on their active user base, and nothing attracts people more than a free product. Social media needs to offer some level of free tier to attract people into the ecosystem.

First, because they are your content creators , people go to social media to see what other people are doing. And once you get them, it will be much easier to promote a subscription service. The second and most important reason is that you can monetize your free users. But what about advertising? All of these companies have advertising and promotional partners who pay based on their reach.

Remember when we said that third-party apps are developed by power users looking for a better experience? What better experience than having no ads in your feed? For example, in the Reddit case, the API filtered these ads so that users don't have to deal with special offers and other interruptions during the browsing experience - which isn't the best idea if you're trying to increase your profitability before opening the capital.

Unfair… for small businesses

Twitter was the first to use paywalls for its API and significantly increase prices with a three-tier system:

  • The free tier only allows 1,500 posts per month
  • The basic level costs $100 per month, aimed at what they call hobbyists. It allows you to post 3,000 tweets per month at the user level or 50,000 tweets per month at the application level. The reading limit is 10,000 tweets.
  • And then there's the enterprise level, which has no price range, but some professionals have estimated it could cost around $42,000 per month for a low use case.

It seems like something is missing, right? Why isn't there a level for students? Or for small projects? The original Twitter used to be one of the most open and welcoming APIs on the market; It was a godsend, in contrast to the obstacles you have to overcome to work with other companies like Meta.

And then there's Reddit. Ironically, the company claimed that its API would charge fair prices, nothing compared to what Twitter was doing. Then, a few weeks later, they made their plans public and all hell broke loose.

Reddit killed the entire app ecosystem that supported it. Apollo, one of the most famous alternatives to the missing Reddit client, had to close because the developer would have to pay more than $20 million per year for an average of 7 billion API requests.

If you think that's a lot, well, it is. Apollo was almost the standard for mobile Reddit. And it was a nice source of income for a small developer. And when Selig had to shut down his app, users were furious, and the situation got so bad that Redditors started a protest against the changes, but to no avail .

Fortunately, the first piece of good news is that no other companies appear to be on the warpath so far. It's still too early to say whether other companies will adopt a similar approach, especially as everyone craves data to continue training and refining their models. This serves as a harsh lesson for startups that want to leverage a service provider's APIs.

Advice for startups

Alright, so you're a startup and you're looking for opportunities to work with APIs, even if they're expensive. What do you do? Here are 5 tips that can point you in the right direction:

  1. We advocate due diligence before investing in an expensive API . This means closely examining the API's functionalities and capabilities and evaluating whether it meets your business needs. A good illustration of this would be a startup looking to harness the power of AI.
  2. We ask startups to consider the scalability of their chosen API . As your startup grows, will the API be able to keep up with increased demand? Will your business still be viable as your user base grows? Contact expert API development services if you are unsure of these answers.
  3. We recommend carrying out a cost-benefit analysis . The expenses of using the API must be balanced with the potential profits and efficiencies your startup can achieve.
  4. We ask entrepreneurs to negotiate terms with the API provider . Many companies have multiple pricing tiers or offer discounts for budding startups. Square's API payment gateway, for example, is known for providing flexible pricing plans for small businesses.
  5. We suggest optimizing your usage to manage costs , especially for APIs that charge per request. Prioritizing necessary requests and saving data traffic can make a significant difference to your overall expenses.

We don't have a Magic 8 Ball to know what the future holds, but for better or worse, these changes seem to be leaving behind one of the best aspects of the internet, a community of smart people with very little budget, but a lot of heart and creativity. Let's hope that in the future we can find a middle ground where everyone wins.

Source: BairesDev

Back to blog

Leave a comment

Please note, comments need to be approved before they are published.