Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: What was the best CS paper you read in 2017?

news.ycombinator.com

Post reply on HN
Print view View original

1–10 of 131 posts

Re: Ask HN: What was the best CS paper you read in 2017?

#2

Post by JelteF »

The raft consensus algorithm article [1] and the follow up that modifies it slightly with some changes that are useful when implementing it [2]. Both are written very clearly and very focused on actually using the algorithm in practice. Problems that arise when implementing it are discussed and solutions are proposed as well.

[1] https://raft.github.io/raft.pdf

[2] http://openlife.cc/system/files/4-modifications-for-Raft-con...

Re: Ask HN: What was the best CS paper you read in 2017?

#4

Post by nickpsecurity »

I read a lot of good ones but I'll answer what was most important rather than best. We're seeing a revival of formal methods ranging from lightweight (TLA+) to heavy. Two problems are in my sight: lack of framework plus consistent, empirical data for evaluating how suitable a given method is for a specific project or company; many who would use lightweight or practical methods will quit because someone convinced them to start with heavy ones. The paper below is a start on the former:

https://www.scch.at/de/rse-news/passende-formale-methon?file...

I think that was a great start on an evalution criteria because its comparisons match what I read in the various experience reports. ASM's and TLA+ coming in the lead on usability but B method on code generation. It will also help to note what projects have been completed or reusable libraries available in each. A person building verified compilers for functional languages might want to learn Isabelle/HOL due to CakeML, doing a new TLS extension learn F* due to miTLS, type-safe assembly learn Coq due to CoqASM, and so on.

For the other issue, I made a proposal on HN first and then on Lobste.rs about avoiding people quitting early from finding formal methods too overwhelming or useless. I proposed having a default intro or lure that matched the goals of different people to tools with a warning of some's difficulty. It got some interesting responses (below) I'll turn into a mini-essay with links eventually.

https://lobste.rs/s/n7v658/your_thoughts_on_this_advice_thos...

Re: Ask HN: What was the best CS paper you read in 2017?

#5

Post by babangida »

It's an old one, but this year I read The Emperor's Old Clothes by C.A.R. Hoare for the first time and it got me started in a Hoare rabbit hole, just a brilliant guy.

I read the annotated version on Fermat's Library - great source of interesting papers (http://fermatslibrary.com/s/the-emperors-old-clothes).

Re: Ask HN: What was the best CS paper you read in 2017?

#9

Post by sulam »

I highly recommend people give this paper a read. I think it points the way to a radical redesign of fundamental parts of the system stack over the next 5-10 years. If you work in systems and you aren’t thinking about this stuff, you’re about to be lapped.

The Case for Learned Index Structures:

https://arxiv.org/pdf/1712.01208v1.pdf

Post reply on HN