The Year in Computer Science

Bill Andrews. Quanta Magazine. December 25, 2024.

Bill Andrews joined Quanta Magazine as senior editor in 2019. Previously, he spent seven years at Discover, where he led the magazine's award-winning math, space and physics coverage. He has also worked at Astronomy magazine and the Mayo Clinic. He received his bachelor's degree in science writing at the Massachusetts Institute of Technology and his master's in journalism at the University of Wisconsin, Madison.

Researchers got a better look at the thoughts of chatbots, amateurs learned exactly how complicated simple systems can be, and quantum computers passed an essential milestone.

The end of 2024 seems a particularly uncertain time in history, and theoretical computer science is no exception. Amid several breakthroughs and new findings, the field also confronted its own doubts and limitations.

For example, artificial intelligence once again dominated the popular discourse this year. Researchers have begun to understand what might be going on within the “black boxes” of neural networks that power chatbots such as Bard and ChatGPT, and to show that these systems truly understand the data they’re manipulating and composing. But there’s also a growing sense that AI’s progress has already started to slow(opens a new tab).

Other areas of computer science saw clearer successes. After decades of hiding, an elusive numerical critter known as the fifth busy beaver finally gave itself up. But even here the news is not all good: Initial searches for its successor — the sixth busy beaver — suggest that it may lie beyond impassable mathematical barriers.

The field of error-correcting codes — mathematical constructions that can fix themselves if an error has occurred — also rose in prominence in 2024. Researchers showed, for the first time, that an error-correcting technique essential for quantum computers really does work. But another finding showed that certain classical error-correcting codes are fundamentally inefficient, dashing hopes for practical versions of this “magical phenomenon.”

Big Beaver Breakthroughs

This summer, a team of enthusiastic amateurs solved one of the biggest open questions in computer science when they found the fifth busy beaver. Named for a particularly industrious kind of computer program that takes a long time to run, the problem is connected to some of the deepest open questions in computer science and mathematics.

The problem involves Turing machines, some of the simplest possible computing devices, first conceived by Alan Turing as a model for a general computer and now easily simulated online(opens a new tab). Depending on their initial settings, these can either run forever or halt after a set number of steps. The busy beaver problem asks: Given only a certain number of rules governing the machine, how long can it run, short of forever? Researchers solved this problem for machines with 1, 2, 3 or 4 rules (1, 6, 21 and 107 steps, respectively), but the fifth busy beaver number eluded them for decades.

Quanta staff writer Ben Brubaker chronicled the efforts of a global team, made up mostly of nonexperts, to find the number — it’s 47,176,870 — and prove that it couldn’t be higher. The problem doesn’t have any direct applications, but finding the solution does represent a kind of victory over overwhelming mathematical complexity — one that early searches for the sixth busy beaver suggest we may never see again.

And in January, we wrote about another diverse team of nonprofessionals who helped solve a different kind of numerical puzzle. This effort focused on the Game of Life (opens a new tab), by John Conway, and the hunt for repeating patterns of specific lengths. By finding patterns that repeated after 19 and 41 steps, they proved that the Game of Life is “omniperiodic” — capable of repeating after every possible number of steps.

Better Living Through Codes

Quantum computers have long tantalized researchers, but despite years

Episode 3658 - December 28 - Tiếng Anh - Điểm nổi bật trong năm 2024 về Khoa học Máy tính - Vina Technology at AI time

Vina Technology at AI time - Công nghệ Việt Nam thời AI

Episode 3658 - December 28 - Tiếng Anh - Điểm nổi bật trong năm 2024 về Khoa học Máy tính - Vina Technology at AI time