Motoko, the language for the Internet Computer
Speakers: Joachim Breitner
Transcript By: Bryan Bishop
Introduction
Thanks, Byron. Glad to be here. My name is Dominic Williams. I am founder of the internet computer project at Dfinity Foundation. Last year we talked about consensus protocols. Today we are talking about tools for building on top of the internet computer which is great news and shows the progress we have made. I’m going to give some introductory context, and then our senior researcher from our languages division will talk about Motoko. Is it working? Alright.
The internet is created by a product called IP. It combines independent networks to create a global public network. But the internet only does connectivity. To create enterprise systems, websites, services and so on, we built on top of a legacy tech stack that is composed of things like cloud, databases, web servers, middleware, application servers, etc. It’s not very secure, of course. Legacy stack also includes things like firewalls and VPNs which don’t work very well which is why even if you patch your software and configure it, often the bad guys will get in. We see hacks, breaches, and now ransomware where bad guys get in and encrypt all your data.
It’s incredibly complex to build systems. This results in high costs and entrepreneurs being less nimble. This is despite the fact that according to Gartner that in 2020 the legacy tech stack will cost the world $3.9 trillion. You’d think that the world would get a better tech stack for $3.9 trillion, but there’s a lot of problems starting with the fact that it’s difficult to create secure systems.
We want to extend the internet so that it becomes the stack. We do this through a product called ICP which combines compute capacity from independent data centers. This adds new capabilities to the internet so that it can act like a giant computer. The idea is that people just write secure internet-native software on to the internet, and the internet now hosts and runs enterprise IT, websites and services. Our proposal is to try to leave the legacy stack behind and build on the internet instead.
What to do with an internet computer
The answer is to build and deploy software canisters which are the compute unit of the internet computer. You deploy these things into a seamless universal software which is secure and scalable. This is tamperproof, unstoppable, it provides for language interoperability, so this means that if I upload some software to the internet computer and I upload some software, written in two different languages… if you share some functions with me and give me permissions to call your functions, then my software can call directly into them ((as if DCOM never happened)). It persists memory pages. We think this is an important advance for a number of reasons. You just write code. Data lives in your code. It’s a fundamentaly different way of persisting data. It supports autonomous software, and it’s fast and scalable. We want you to use our internet computer to build the next generation of websites, enterprise services, platforms, smart contracts, and it’s worth saying that canisters runs on all the security guarantees of the smart contracts so you can use canisters as smart contracts if you want.
I mentioned open internet services. A month ago, we demonstrated the bronze release of the internet computer network which is currently in incubation at Swiss data centers and we did this using an open internet service called Linkedup which is like Linkedin the professional business profile network. The computer dynamically created user experiences directly into browsers in just a few milliseconds. This is one of the pretty cool things about internet computers. It’s fast. In many cases, it will be faster than today’s architecture where you have hyper-scale data centers and CDNs. The internet computer is a native edge architecture and it dynamically creates content for you on the edge of the network.
Blockchain
The internet computer is a blockchain computer. Canisters are roughly analogous to smart contracts. With that, I will hand you over to Joachim to talk about Motoko which is a language specialized to develop canisters.
Motoko
Thank you for that intro. The technical aspect is that we want language and durability. These canisters that he talked about are webassembly programs. This is a target for many languages out there like rust and C. Why do I have to learn about a new language if I want to use the internet computer? The answer is that you don’t have to, but you might still want to because by designing a language tailored to the internet computer platform then you reep certain benefits.
Motoko design goals
The design goals are easy adoption, it should be easy to use and avoid foot-guns, and it should be seamless and it should integrate seamlessly into the platform. For the rest of the presentation, I am going to show you real code and run it and try to explain to you what I mean by familiar, safe and seamless.
Demo
This is the first piece of code I am going to show you. I don’t think I have to explain to you what this code does, precisely because this language is familiar. Based on my background you might expect pure functional programming, but no we opted for something more familiar. This gets compiled into web assembly by running:
moc -wasi-system-api factorial.mo
I can run this in the wasmtime web assembly runtime and it gives me a result. This code shows a few of the safety-by-default decisions we put into Motoko. One of the design decisions is that you have to explicitly say when you want a variable to mutable. This already helps for a few bugs.
Another safety feature is one you don’t see in the code; but in the last talk there was a discussion about overflow. It’s a problem if it happens on-chain and some positive goes to negative or something. By default, Motoko uses unbounded integers so they simply can’t be overflowed. This design decision is a tradeoff. Integer overflows have known problems. Memory consumpation of big integers can vary a lot, although this isn’t a new problem. You have the same problem every time you read a string from a user.
The next building block to make this language safe is good expressive data structures, and a type system to support them. For the next example, first we see that we define a value, it’s a record that represents a customer. We have records, strings, tags, these hash things… What I am going to point out first is that you don’t have to mention any types about defining this value.
Sponsorship: These transcripts are sponsored by Blockchain Commons.
Disclaimer: These are unpaid transcriptions, performed in real-time and in-person during the actual source presentation. Due to personal time constraints they are usually not reviewed against the source material once published. Errors are possible. If the original author/speaker or anyone else finds errors of substance, please email me at kanzure@gmail.com for corrections or contribute online via github/git. I sometimes add annotations to the transcription text. These will always be denoted by a standard editor’s note in parenthesis brackets ((like this)), or in a numbered footnote. I welcome feedback and discussion of these as well.
Tweet: Transcript: “Motoko, the language for the Internet Computer” https://diyhpl.us/wiki/transcripts/stanford-blockchain-conference/2020/motoko-language/ @nomeata @CBRStanford #SBC20