HNHacker News
TopNewBestAskShowJobs

drongoking

376 karma · joined August 26, 2016

submissionscomments
drongoking··on The Rise of Peer Review Bots
The author compares it to YouTube but I don't think it's quite that bad. I doubt we'll see the relentless adapt-and-attack cycle that YouTube gets. The goal of researchers has always been to publish, but not every venue is equally reputable and the community knows that. Researchers who publish lots of weak papers in crappy journals get a reputation. Tenure committees know how to judge a publication list.

As for bots, journals need to allow authors to respond to bots' rejections and have the rebuttal go to a human editor. True plagiarists won't bother, but if a bot is just catching superficial content this should be easy for an author to point out. Once editors get tired of dealing with false positives, the bots should get fixed.

(As for p-values, that's a larger issue that's been discussed for a long time.)

drongoking··on The end of capitalism has begun (2015)
This article is from 2015. Is the beginning of the ending of capitalism done beginning?
drongoking··on Literature-Based Discovery
Within a field it's what most scientists do. Between fields, or subfields, it's less common. When a field gets very large such that no one can keep up with it all, Literature-Based Discovery by machine is useful.
drongoking··on Landfill is underrated and recycling overrated
As a meta observation, he seems to assume that materials production and disposal would improve over time while recycling technology would not, and that the economics wouldn't change either. Some of the sources he cites for the inefficiency of recycling date from the 1990s and he's obliged to find something more recent.

Arguably recycling technology has more potential to improve since large-scale recycling efforts in the US have only been around since the 1980s. Landfills have been around forever.

drongoking··on Tips for Writing a Technical Book
Amen to this. In fact, I tell people that if they're interested in writing a book they shouldn't be asking themselves "Should I write a book?", they should be thinking: "I have 18 months of free time. What could I do with it?"

That's a lot of free time. You could:

- Learn a musical instrument pretty well. - Learn a sport and get pretty good at it. - Become a major committer on a large open source project. - Pick up a foreign language and achieve proficiency. - Make a lot of money on the side. - Go back to school and get a degree. - Concentrate on your social life. - Write a book.

If after considering these, you decide that writing a book is the best use of your time, go for it. Think of the money as a nice bonus.

drongoking··on Tips for Writing a Technical Book
Speaking as another O'Reilly author, I'd agree with this and add that for tech books the publisher is even less necessary. If you've written papers with Latex/MD/asciidoc, and drawn figures and tables, you're already doing most of what you'd do with your publisher's toolchain anyway. Hook up with Amazon's on-demand publishing service and (except for the nice cover) you've got it almost end-to-end.

[I assume you can write and organize thoughts decently well, and have some people who can proofread for you. If not, a publisher really won't help too much with that anyway.]

So what does a publisher provide? Marketing, promotion, distribution, reputation. No matter how good of a book you write, having a name publisher like O'Reilly gives you substantial credibility that nothing else will. They can arrange promotions through conferences and email distribution that you probably can't. I wouldn't underestimate this, but I also wouldn't necessarily think it's worth their 80%-90% cut either.

drongoking··on Machine Learning: Full-Text Search in JavaScript – Relevance Scoring (2015)
Your distinction between TFIDF as simply computed vs embeddings as learned is odd and artificial. Both are computations from data, but TFIDF has an understandable closed form and word embeddings do not. As for machine learning, it has to do with improvement and doesn't even necessarily need data.
← PreviousPage 3 of 3