The Future of Software Development

TL;DR

The way we think about computation and software development is changing due to advancements in big data, distributed systems, and the need for simplicity.

Transcript

all right so I'm going to talk a little bit more about the software side of things right and it's kind of interesting the way this played out in that I think my talk is actually going to bring a lot of this sort of discussion we've had so far together into something interesting and and so you know we talked about these major trends in computing thi... Read More

Key Insights

• 🫵 The traditional view of a computer has evolved, with machines becoming larger, networked, and incorporating diverse devices like thermostats.
• 🚗 Performance is crucial as the computing landscape shifts to mobile devices and energy efficiency becomes paramount.
• 👶 Industry and research are coming together to find solutions to the challenges posed by the new machine.
• 🥺 Simplifying software development by embracing foundational principles and simplicity can lead to better outcomes.
• 😆 Innovations like CRDTs, immediate mode UIs, and constraint programming are examples of how simplicity is being pursued in different domains.
• ❓ Democratizing computation becomes more possible as programming becomes more simple and accessible.
• 📚 React, a popular JavaScript library, has already incorporated CRDT data types, indicating the adoption of simpler approaches in industry.

Questions & Answers

Q: Why is performance becoming increasingly important in computing?

With the rise of mobile devices and networked systems, squeezing maximum power and efficiency out of small devices has become crucial. Waiting for faster hardware is no longer a viable solution.

Q: How are industry and research coming together in the field of computing?

After a period of separation, big corporations driving the agenda, and skepticism towards computer research, the industry is turning back to research for innovative solutions and foundational principles.

Q: What are commutative replicated data types (CRDTs) in distributed systems?

CRDTs provide a simpler approach to coordination in distributed systems, allowing each node to update information regardless of order. This sidesteps the complexity and difficulty in reasoning about traditional coordination mechanisms.

Q: How are immediate mode UIs simplifying user interface development?

Immediate mode UIs involve redrawing the entire interface every frame, making it easier to reason about and achieve better performance. It simplifies UI development, especially for complex and fast-changing interfaces.

Summary

In this video, the speaker discusses the changing landscape of software development and computing. They emphasize the importance of simplicity and the need to find new ways of thinking about computation in order to adapt to the new machines and advancements in technology. They highlight some interesting examples, such as commutative replicated data types, immediate mode UI, constraint programming, and the re-emergence of relational databases without SQL. The speaker also suggests that simplifying programming could potentially democratize computation and make it more accessible to everyone.

Questions & Answers

Q: What are some of the major trends in computing that the speaker discusses?

The speaker mentions big data, deep learning, and mobile as some of the major trends in computing.

Q: How has the notion of a computer changed in recent years?

Over the past 15 years or so, the traditional idea of a computer has changed. It is no longer just a standalone machine, but rather a networked system that includes not only large computers with thousands of cores and huge amounts of memory and storage, but also smaller devices such as thermostats and laptops.

Q: How does this change in the notion of a computer affect software development?

With the shift towards networked systems and the increasing importance of performance and efficiency, traditional models of computation no longer fit. Developers struggle to write efficient programs that work well on these new machines. This has led to a convergence of industry and research as they seek new ways of thinking about software development.

Q: How does the speaker define simplicity in software development?

The speaker defines simplicity as the ability to gain power through simplicity, rather than adding more complexity to software. They argue for finding foundational principles that can help embrace the new machine and make computation simpler, rather than adding more layers of abstraction.

Q: What are some examples of new approaches in distributed computing?

One interesting approach in distributed computing is commutative replicated data types (CRDTs), which allow updates to happen regardless of order. This sidesteps the problem of coordination in distributed systems, making it simpler and easier to reason about. Additionally, systems based on append-only event logs are being developed, where every event is logged and used to determine the state of the system at any given time.

Q: How is simplicity being applied to user interface development?

In the realm of user interface development, the speaker mentions the emergence of immediate mode UI, where the UI is redrawn every frame based on the state of the application. This approach simplifies UI development, making it easier to reason about and faster to perform.

Q: How is constraint programming making a comeback?

Constraint programming, using tools such as SAT solvers and SMT solvers, has become significantly faster and is being applied in various areas. For example, Apple's Auto Layout in iOS is based on a constraint solver called cassowary. Constraint programming offers a simpler way to solve problems by setting up a set of constraints for the system to solve.

Q: What is the speaker's view on relational databases and SQL?

The speaker mentions a group of people who are exploring the idea of getting rid of SQL and bringing back the relational database. They question whether a general-purpose database can compete with specialized ones in terms of performance. The re-emergence of data log, a language for relational databases without SQL, is an interesting development in this area.

Q: What is the potential impact of simplifying programming on democratizing computation?

The speaker suggests that simplifying programming could open up opportunities to democratize computation. Currently, programming is seen as difficult, but if the system can be simplified and made more accessible, more people would have the ability to engage in simple computation. This could have significant implications for accessibility to technology and the democratization of knowledge.

Takeaways

The speaker believes that simplicity and a return to foundational principles are important for adapting to the changing landscape of computing. They highlight the convergence of industry and research in finding new ways of thinking about software development. Examples such as commutative replicated data types, immediate mode UI, constraint programming, and the re-emergence of relational databases without SQL showcase the ongoing efforts to simplify and innovate in the field. Simplifying programming could potentially democratize computation and make it more accessible to everyone, with significant implications for the future of technology.

Summary & Key Takeaways

• Computing trends like big data, deep learning, and mobile have been discussed, but there are underlying changes that are often overlooked.

• The traditional notion of a computer has evolved, encompassing more than just a physical device.

• Performance, efficiency, and networked systems have become crucial considerations, challenging traditional models.

• Industry and research are converging to find solutions, embracing simplicity and foundational principles.