We built NotRealTwitter as a fun demo, but what does it mean for the future of software development? Can real Twitter actually be built without code?
In the past few years “no-code” has become one of the buzziest terms in tech and no-code companies one of the hottest investments. In the past year Webflow raised $140M, Unqork $207M, and Outsystems $150M, all with the promise of democratizing software creation and getting rid of the coder bottleneck.
But there are plenty of skeptics, and no bigger ones than software engineers. Ask most engineers whether no-code can replace code and they’ll scoff and talk about how their work requires too much precise control, something that no-code tools don’t allow by their very nature.
So what’s the truth? Does no-code represent a generational shift in how software is built and coders will find themselves out of a job? Or are these tools more of a toy and will be forgotten in 10 years?
We think it’s a generational shift. But, at the same time, we believe that the truth is more nuanced - that most (i.e. largest number) of software applications will be built without code but that the most complex applications will continue to be built with code.
To better understand this prediction let’s talk about easy problems and hard problems.
In software development easy problems are those that have been solved lots of times before and, as a result, there exists a standardized way of solving them. For example building a two-sided marketplace (e.g. Airbnb for x) is most of the time an easy problem. Under the hood there is a table of users, properties, bookings, etc. When a new booking is made, the application makes an API call to a payment service like Stripe to charge the user and then a new booking object is created. None of this functionality is particularly hard and, because of that, it can be abstracted and be built without code.
On the other hand, you have hard problems. That same Airbnb for x platform might be generally simple but might also have some complex elements, such as:
- A machine learning algorithm that surfaces the properties that are best matched to users based on their previous bookings
- Pages optimized to load in under a second across millions of page loads per hour
- Integrations with complex legacy booking systems
The reason why these problems are hard is because we don’t yet have standardized solutions for them, and thus we have to solve them in a very precise & unique way, using code.
The thing is that the large majority of custom software being built solves easy problems. And because of this, we believe that most such software is going to be built without code. At the same time, we also believe that hardcore coders will have plenty of work solving the hard problems. And then there will be a blend of the two - no-code solutions that accomplish the 90% of the easy functionality integrating with coded solutions that accomplish the hard 10%.
So, should Jack Dorsey consider moving the real Twitter to a no-code stack? In our opinion, no. The core functionality can certainly be built in that way, as proven by Not Real Twitter. But the dealbreaker today is scale — supporting nearly 200 million users and 6,000 tweets every second creates a unique complexity that clearly constitutes a “hard” technical problem. But should an early-stage social network consider no-code for their development? Absolutely.
Want to learn how to build web apps without code? Check out our free no-code bootcamp!