[_gabe_]

I’m sorry, what? The best computer scientists I know, Djikstra, Tony Hoare, Turing, Knuth, and more, all developed practical algorithms and concepts that are still used today! Like the parent commenter said, it’s really not difficult to provide code that works and supports your conclusion. I’ve also read computer science papers that bury the lede. It’s apparent when you try to replicate them that the author was disappointed that the work didn’t support their hypothesis, so they made it look good. It’s sad because a true scientific pursuit finds the knowledge valuable whether it works or not! They just need to state it instead of hide that fact.

Your whole comment reads as some sort of weird gate-keeping where people without the “proper” education could never fully understand a “true” paper. We can, and we do. There’s a reason we learn from truly great computer scientists like I listed above in college, and not the people posting unreproducible work.

[jltsiren]

Early computer scientists discovered many fundamental ideas. Since then, the field has become more specialized and technical.

My comment was not gatekeeping and more in line with "if you are a mechanical engineer, don't expect to get much value from papers in theoretical physics". "Proper" education or otherwise spending a few years focused on the topic may help, but it's far from foolproof.

The "if you start investigating where that idea came from" part came from my personal experiences. There have been times when I've categorized a paper, on a topic I'm supposed to be an expert, as interesting but practically irrelevant. Only to later find that it became a building block for a major practical result.

If you read a research paper and think you understood it well enough to judge it, you are probably wrong. Even if you are an expert on the topic.

[withinboredom]

> If you read a research paper and think you understood it well enough to judge it, you are probably wrong. Even if you are an expert on the topi

That sounds like utter nonsense to me, but I go through about 5-10 papers per year.

> There have been times when I've categorized a paper, on a topic I'm supposed to be an expert, as interesting but practically irrelevant. Only to later find that it became a building block for a major practical result.

This is usually why I am investigating a paper. I want to understand what they did because I see how it can be used practically, if what they wrote is correct.

I'm working on the wpaxos paper right now. There's even some example code, totally missing some aspects of the paxos algorithm (it also doesn't appear to run in its original form). There's a whole section in the paper about transactions, marked as "//todo" in the code. The flexible quorum implementation is subtly incorrect as well. Anyway, it's a pretty good paper, and I'm fairly certain a large part of it is probably driving cosmosdb (since they work at Microsoft now and I recognize a number of possibilities that echo what cosmosdb can do, along with some same constraints and it would explain the weird billing; but this is a total guess and I have no proof of it). Anyway, your comment is something I totally agree with. Most people lack the required knowledge, creativity, and insight into most papers to truly understand what the value is in the research.

For fun, I am researching magnetic resonance through an iphone's magnetometer to detect stalkers. If it works, I'll have an app (assuming apple doesn't have any issues with it). I wish I could write a paper on it, but I am nowhere near academia these days; which is a whole different shit the academics have done to themselves to make it so only academics can contribute to public knowledge research.

The whole publishing papers thing is broken IMHO, and no, I don't have a solution.

[jltsiren]

When a paper is published, there are usually approximately 0 people in the world who really understand it. That number includes the authors. But if the authors continue working on the same topic, they should understand it much better after a few years.

Research papers are published early in the research process. Typically when there are some interesting, likely correct, and possibly relevant results to show, but long before anyone manages to put the results in the proper context.