Hacker News Reader

Comment by bregma | Hacker News Reader

It would probably benefit society as a whole if the abstracts of scientific papers (not their content) eschewed sesquipedalian loquaciousness.

Here's the same abstract for the more nearly human.

  This paper presents evidence and analysis that shows large magma chambers can be found under all non-extinct volcanoes on the west coast of North America.

View on HN

Idk. You can always just leave out details to make things simpler. I could rewrite your version of the abstract to "Scientist find magma under volcanoes." Much simpler! Leaves out a lot from yours, but so did yours leave out a lot from theirs.

I think in this case it is important to consider the audience one is writing to. And the abstract is written to other geologist. They want to know what kind of evidence they have (purely computational? new field measurements? what kind of modality?)

While for me this sentence[1] is hard to penetrate jargon, presumably for the intended audience it is packed full of information. Same as the sentence "we persist the setting in a local sqlite database using sqlalchemy ORM" means a lot more to some dev like me than "we save the setting". You can simplify it of course, but while you make it easier to understand to a lay audience it becomes less specific to the intended and specialist audience.

1: "seismic imaging of subvolcanic magmatic systems along the Cascade Range arc from systematically modelling the three-dimensional scattered wavefield of teleseismic body waves"

I disagree. This abstract is a mouthful, no doubt, but it's a highly information-dense mouthful, and it does its job well.

Along the spectrum between intra-disciplinary academic communication on one end and public-sphere science communication on the other, the abstract is and should be just above the level of the paper it's abstracting. It's meant to be a technical summary for academics, and should not be dialed down for the sake of non-technical consumption. This one - speaking as a geologist with a long academic history - reads quite well, I think.

In contrast, the sentence you propose should be the output of responsible and effective scientific communicators working closer to the public-sphere end of the spectrum, say, at the level of phys.org, here on hackernews, etc.

Many journals use a separate summary statement which describes the findings and significance in more ordinary terms (eLife is a good example). An abstract has a specific purpose and it is not quite meant for a public audience. It can also vary depending on the journal, with less specialized journals being written for broader audiences.

(Although I will agree this one is a bit unnecessarily verbose)

That's squashing it too much. The angle that magma chambers could be diagnostic of eruption risk is what I think the public would find most interesting. The results seem to say that it isn't so.

Obviously that fails at actually being an abstract: it doesn't communicate what the novel research is, or anything about the methods.

Journals aren't for lay people. It's the pop science articles that should do better.

Also, honestly, this abstract was much more comprehensible than I expected.

That is what the press release from the university should have, but many of them add a lot of hype and distort the paper. Something like

fake> For fist time, scientists of $OUR_UNIVERSITY have seen Doors of Hell with their own eyes and are going to use it to replace Nuclear Power in 5 years and solve Climate Change and find the Atlantis.

I'm exaggeration a little, but I've seen a few horrible cases https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...

I wish they would have both. Personally when I'm reading something in my field, I want the specific details, but when I'm reading outside my field, I could use the dumbed down version.

What sort of evidence and analysis? Did the scientists dig a tunnel underneath these volcanos and drop a dipstick? Point being that the “loquacious” information you removed from the summary contained useful information and your summary just tosses it.

This is a problem I expect to be fully and thoroughly solved by LLMs very soon; no longer should we have to concern ourselves with rephrasing things for different audiences than our peer groups. Scientists often aren’t also great writers, and if they don’t have to be, they can focus more on their actual area of expertise.

It boggles my mind that there are 100 products for AI summarization for every 1 that does filtering, classification, recommendation, clustering, etc.

It seems some A.I. influencers can ask for a kilogram of antimatter and get it, but when the rest of use use Google, we get worthless A.I. summaries at the top of the page, like the one that told me to reset my network switch by (1) turning it off, and (2) then pressing the reset button for five seconds (does nothing because... no power!)

This study shows the kind of results that the "rest of us" get

https://www.theregister.com/2025/02/12/bbc_ai_news_accuracy/

Here is a good AI tool that you can use for free.

https://cofyt.app for YouTube summaries, instant takeaways, and chat with video transcript.

Don't complain, try it and come back here and give your feedback so that others can read.

That's a better use case. I can skim a long document quickly but I can't quickly skim a 30 minute YouTube video. It's pretty fast though... Does YouTube already have a transcript that it's reading?

I'm not suggesting in any way that this replaces watching all YouTube videos but it could help you save some time with some types of video content and a co-pilot for others.

You can try to ask it for things to your requirements,be descriptive.

For example: - you can ask it to list topics or tags in that video.

- you can ask it for specific timecodes/ Timestamps when topic X is mentioned

- you can ask it to tell you what's happening in time blocks/Chunks of x minutes.

be descriptive and ask for the format you want: lists, bullets, list with description, etc

Thanks for the feedback.

While I agree with you, note those directions may not be wrong. Many modern electronic devices do not completely power down when you turn them "off". Unplugging does (once the capacitors drain) but not necessarily using the power switch.

For this device that's wrong

https://store.ui.com/us/en/products/usw-lite-8-poe

but I would say that device is poorly documented, you have to go to the forums to get all details, one of which is that you definitely don't want to hold down the reset button when it is off and then turn it on (holding the button down) because that will boot it up in TFTP mode which I guess loads the system software over the network.

Mine wasn't lighting up any lights when I plugged ethernet cables into it, the reset button didn't do anything if the power was on or off. They RMAed it, I hope the next one works out better.

Reply on news.ycombinator.com