Streaming browsers and the future of web apps

1/ As technology, standards, and incentives evolve, all software applications will eventually move to the web -- with the browser serving as the operating system. We are already almost there today. But there are some challenges.

2/ Software today is written in 2 parts - code that runs on your client (the browser), and code that runs on the cloud (the servers). Because of the way the internet has evolved, a lot of effort is spent in linking the two pieces together, since they're effectively mirroring the data model and functionality.

3/ As client side code has gotten more complex (imagine gmail or figma), it has pushed browsers to do more, and consume more resources on the local computer. And with the fragmentation of web applications, the number of tabs in use has ballooned too. It also means that more and more native OS functionality (like accessing the web cam) will eventually be available to the browser.

4/ What if we we could create a system where this code only had to be written once, on the cloud and streamed, much like how Netflix streams movies? (Assuming of course that it had parity with all the features we expect of such applications? It would have a ton of benefits.)

4a/ The browser would simply become a terminal used to access remote services, and perhaps start to serve more as a container of user data and a way to grant online applications permissions to use it, vs having them collect it by default.

5/ Code would only have to be written once, and deployed without worrying about the client side. This would make things much more secure and packaged. It would mean that app stores would not be as relevant in the long term, since every application would be available simply via a URL.

6/ Since there would be barely (if) any client side code, each website would be very secure, removing the spectre of cross side scripting attacks.

7/ Imagine needing to add on a background image to your camera feed - rather than doing it on the browser like today, the video would simply be sent to one location, filtered, and streamed to whoever needed it without causing any more resource usage on your laptop.

8/ Machine learning models that are today run in large cloud based systems could be run specifically for each user, in their own silos, without worrying about end user hardware capacity, or data privacy issues.

9/ Rather than building hacky client side integrations of other applications via chrome extensions, real APIs could be built that all run in clean silos and with access controls directly on the cloud. This means that integration between disparate apps can become much easier and help reduce some of the fragmentation that has crept up on us.

10/ All your content can be live indexed in your own silos - a whole new generation of search engines can come up where they request permission to index your data, putting you in control. They can then run powerful algorithms to give you a much more pleasant search experience that is integrated into your daily workflows.

11/ @suhail's @mightyapp that launched last week offers a peek into such a future. It's an impressive technical achievement, rooted not only in solving a real user need today, but shows that a future like this is only constrained by time.

11/ In the same way that NVIDIA CUDA like SDKs allow us to offload significant mathematical computation to dedicated GPUs, I expect @mightyapp to start offering libraries where developers can offload significant computing to the Mighty cloud.

12/ Since Mighty is centralized but does not own your data, their SDK in the future could complement OAuth and other identity systems to give more granular control of user data to applications.

13/ A final thought on pricing. While there is a lot of skepticism around a 30-50$ a month price point, the analog in my mind is this. Photoshop users invest in more RAM, a better Wacom tablet, better GPUs and other accessories to get their work done. These resources are fixed costs and need to be upgraded at regular intervals given the rate of innovation. Power users of web applications on the other hand don't have such resources available to them - the best they can do is RAM or a better laptop, but the browser is still a fundamental bottleneck. MightyApp offers them a way to get constant ugprades on all the things that matter, and prices itself in a SaaS model rather than a fixed "licensing" model. Over time, I expect this price to drop as economies of scale and efficiency start to play in the mix.