Jonathan Blow: "C++ is a weird mess"

The Witness developer on his new programming language that could increase productivity by 50 to 80 per cent

Sometime around 2014, while still developing The Witness, Jonathan Blow sat down to work and realised what a miserable experience programming could be.

Why was that? Well, as Blow says, it's because C++ is a "really terrible, terrible language." During his talk at Gamelab Barcelona last week, Blow laid out what he believes are the issues with the commonly used language, and what he hopes to achieve with the development of Jai, his self-made alternative.

The main flaw with C++, in Blow's opinion, is that it's a fiendishly complex and layered ecosystem that has becoming increasingly convoluted in its effort to solve different problems; the more layers, the higher the stack, the more wobbly it becomes, and the harder it is to understand.


Jonathan Blow

"People think a programming language should be a complicated ecosystem," Blow said to the audience. "I don't want to learn an ecosystem for a year, I just want to put 3D graphics on there."

With Jai, Blow hopes to achieve three things: improve the quality of life for the programmer because "we shouldn't be miserable like many of us are"; simplify the systems; and increase expressive power by allowing programmers to build a large amount of functionality with a small amount of code.

Currently, Jai boasts an improved productivity of around 15 per cent, but Blow believes that, given time, it could be anywhere between 50 and 80 per cent. caught up with Blow following his talk to learn more about Jai which, if his predictions for the language are accurate, could reshape game development at both the indie and AAA level.

According to Blow, "C++ is a weird mess". It began life as C in the 1970s and changed substantially over the next decade or so, layering on "historical cruft and weirdness" that holds back development with the need to remain compatible with historic decisions.

"There is a very large status quo bias around things like this. We make games a certain way; people make all these other programming languages and yet we don't use them"

"Then C++ is a language where they said: 'We're going to be C with a bunch of stuff added' and that was an okay idea. But they made a lot of really bad decisions in the beginning and then they had to repair a lot of those decisions later on, and it's kind of a mess," says Blow.

"If you come to it and you're trying to learn it, a lot of what you're learning is just this rubble... when really you shouldn't have to deal with that at all. C++ is a powerful language in some ways, because it's what we make games in so obviously it's doing something right. But it makes it a lot harder than it should be."

As a project gets bigger the longer it takes to make changes, Blow explains, and the nature of programming means that significant amounts of time get dedicated to "what is essentially busywork."

"I just started thinking: 'Somebody should do something about this because it's ridiculous'. And I realised that nobody was going to do anything," he says. "There is a very large status quo bias around things like this. We make games a certain way; people make all these other programming languages and yet we don't use them.

"Companies have very large game engines that would be very costly to replace. So a game like Assassin's Creed, last I heard a couple of years ago, was like ten million lines of source code... It's probably bigger now."

This "status quo bias" creates inertia within the industry, Blow argues, and big developers aren't going to jump at the chance of rewriting all of their systems when they already have one that gets the job done, irrespective of how cumbersome that process may be.

"I just started thinking: 'Somebody should do something about this because it's ridiculous'. And I realised that nobody was going to do anything"

"So are they going to just turn around tomorrow and just rewrite all that?" says Blow. "No they're not because it would be almost impossible. So a lot of people have this defeatist attitude of 'this can't ever be different than it is today'. I actually had that attitude also. That little realism voice was in my ear telling me I couldn't change this. But then for some reason I decided to question this: how true is that really?

"The more I questioned it, the more I said 'actually this is quite changeable. I think I see a specific thing we do here'. And as soon as I had the idea that there was a way to change it, I was suddenly a lot happier about my day-to-day work because suddenly there was a way out."

By this point, Blow and his team were deep into developing The Witness, so Jai remained a side project. Several years later, however, it's the primary focus of Blow and his team at Thekla Inc. His currently unannounced game project, which he teased during Gamelab, is built using Jai in an effort to "test the ideas against reality."

In many ways, Blow is uniquely positioned to develop a new programming language. He's independent of large studios and publishers, has worked on extremely long projects (The Witness took around seven years all told), and is an experienced programmer.

His history as a designer is also crucial, he says, to making a language that works for building games in the hopes of avoiding the many pitfalls of C++. Even so, Blow is aware that Jai faces the same danger of becoming a "pile of rubble".

"It is a danger, yes," he says. "You could spend all this time trying to design something that you think is better and end up with something that is same or even worse. That could happen. But that's where my experience as a designer comes in. I've designed games that are big and complicated, and you can design a programming language the same, with an eye to the aesthetics.

"I don't think it's too controversial to say that most of the time, when programming languages are designed they are not designed by what we think of as designers"

"First of all trying to keep it simple, but secondly when you do add complexity -- because you always need to in order to make things happen -- how much benefit is coming from that complexity? How easy something is to understand [for the player] is what we think about in game design all the time. If somebody walks into this room... are they going to understand what is going on? How do we help them understand?

"Those same skills can translate to programming language design. I don't think it's too controversial to say that, most of the time, when programming languages are designed they are not designed by what we think of as designers. They are designed by technical people who are trying to solve a technical problem. So the hope is that bringing some design experience in there helps us get to a better result. And it seems to be working. I'm pretty happy with where we are. It's just a lot of work to get it done."

Ultimately, Blow hopes that Jai will be powerful enough to program the biggest AAA games, but right now he is working on an "intermediate step" with his new game.

"It's a real game in all respects, we just picked a design that was a little easier to do. Once that is done and it works and we ship it, then we can scale up that graphics engine and so forth."

While there is no end in sight for Jai, Blow hopes to have a closed-beta out by the end of the year. Whether it will live up to its potential, or his personal hopes, is a long way from being decided. is a media partner of the Gamelab conference. We have attended the show with the assistance of the organiser.

More stories

Thekla doubles grant money for underrepresented devs to $100k

RAD Game Tools' Jeff Roberts matched the initial $50k pool available to indie puzzle designers

By Matthew Handrahan

The Witness studio offering $20,000 grants to underrepresented developers

Jonathan Blow's team keen to see more women, people of colour, disabled people and LGBTQIA making puzzle games

By James Batchelor

Latest comments (5)

Andrew Jakobs Lead Programmer 3 years ago
oh god, not yet another language nobody was asking for except the people creating it..
There are so many languages, and mostly because people didn't like the languages they were using, and yet most also do exactly what the others do, except in a lot of cases the syntax..
And in reality all he's doing is creating another game-engine/framework if I understand what he's doing..
Most AAA engines already use some sort of scripting language like C# which get compiled to native code for a specific platform.
This is just yet another attempt to re-invent the wheel for which are already perfect other wheels available..
5Sign inorRegisterto rate and reply
Omar C.3 years ago
And in reality all he's doing is creating another game-engine/framework if I understand what he's doing..
He is not, he is creating a new language expected to be as low-level as C or C++ in term of control and performances (no GC etc.) but also vastly more flexible and modern. There are large amount of material on his youtube channel to check.

It's ok if you are content with existing solution, he's not creating the language for you or me for but himself first. At least he is taking a different angle at trying to solve the fact that "C++ is a weird mess" (which is not even remotely controversial). The point of his talk - which I haven't seen - appears to be that if every one gives up and no one tries to aim at the moon we'll never get out of our C++ hole. It's nice that some are trying.

Edited 3 times. Last edit by Omar C. on 3rd July 2018 5:38pm

5Sign inorRegisterto rate and reply
I understand the fun/hubris of creating a new language but Haxe is already there. He could help the haxe team.
0Sign inorRegisterto rate and reply
Show all comments (5)
Keldon Alleyne Strategic Keyboard Basher, Avasopht Development3 years ago
@Lionel, but does Haxe meet his actual needs?

One of the biggest issues for C++ is the lack of modules, which they're likely to address for C++20.

Delphi can compile 1 million lines of code in 5 seconds. Such blistering compile speeds would give developers a massive productivity boost.
1Sign inorRegisterto rate and reply
Alex Barnfield Lead Engineer, 17-BIT3 years ago
The history of C++ isn't that relevant, no more than the how hungarian notation came about. The only issue is why is it adopted now.

The philosophy behind C++ - only pay for what you use. The stitching together of features isn't entirely accidental and a well designed engine encourages users of it to only use a subset of the language in a certain way, the consistent vision should come at the engine level.

I was lucky enough to help design an engine at my last job, and most of the gameplay code was done with a variant of the ECS model

All in cpp, no headers for systems, no forward declarations, no complex hierarchies. So many of the issues listed in the JaiPrimer just don't strike me as unavoidable. They are issues, but I've always thought of them as an issue of poor code design than intrinsic problems with the language.

Many other complaints just feel so trivial, every engine typedefs sized variables, std::array is available if you want to maintain the size information, and if I need to change a function signature that is what the VA refactor menu is for. Not that tackling these issues isn't worthwhile, but going on the offensive over them feels like the wrong approach.

I personally think the criticisms of C++ are just distracting. It'd be more productive to move the focus onto the merits of Jai.

Edited 2 times. Last edit by Alex Barnfield on 5th July 2018 4:35am

1Sign inorRegisterto rate and reply

Sign in to contribute

Need an account? Register now.