How to read papers

[studiot]

We frequently recommend papers.

Here is a good book on how to extract useful information from them.

I know it was produced for the medical sciences but the lessons are universal.

A further tip for students.
If you know how the Professors are reading the papers before marking them,  it helps in their preparation.

 

 

 

 

 

[Trurl]

I find academic papers hard to read. I know they are structured for a reason, but why is a Wikipedia paper way easier to understand. What if we turned in our papers in Wikipedia format. Need to get background information click the picture to bring up a supporting article.

IMHO the idea behind the paper is the most important part. Explaining it clearly is the second. I may be simple in preferring Wikipedia but are you saying that I am missing out on current research because I don’t understand journal entries?

But what if you had an abstract and submitted an entire science notebook? With a program such as Mathematica you could link notes and ideas and maybe still have a research paper as a roadmap and the process of your research would be complete.

You would be less concerned in what you did or did not include in the final paper and more concerned on the overall presentation of the cumulating of notes, brainstorms, equations, trail and errors, etc.

[CharonY]

Papers have a different purpose than wiki articles. The latter are oftentimes superficial summaries, which, in the best case also include some synthesis from multiple sources. For many articles this is not the case, however.

The structure of a paper allows someone within the field to evaluate the hypotheses, assess the quality of the approach and possible weaknesses (methods section), the quality of the quality to support their hypotheses (results section). Discussions and introduction not only provide overall context, but also tells the readers where the researchers are coming from, and allow experts also to evaluate overall conceptual weaknesses and strengths. A paper is essentially a discussion tool for scientists.

Changing that to a wiki article would defeat its overall purpose. I think you are not missing out much if you do not understand papers, it just means that you likely lack the necessary training to get the information out of them. After all it is a specialized tool, which is quite the opposite to an encyclopedia like wiki, where accessibility is more important than depth of information.

13 hours ago, Trurl said:

But what if you had an abstract and submitted an entire science notebook? With a program such as Mathematica you could link notes and ideas and maybe still have a research paper as a roadmap and the process of your research would be complete.

That is a horrible idea. A paper is a cleaned up synthesis of often a lot of work. Looking at the raw version of it, would be entirely incomprehensible to except for the person generating int. I have to keep telling my students to clean up their notes as often I barely understand what they mean and over time, they will also forget what their notes mean. Research is almost never linear and 80-90% of the material never makes it into a paper for good reasons.

 

13 hours ago, Trurl said:

You would be less concerned in what you did or did not include in the final paper and more concerned on the overall presentation of the cumulating of notes, brainstorms, equations, trail and errors, etc.

This again is the opposite of what science is about. Science is about the synthesis of outcomes not just a bunch of data aggregates.

I am repeating myself, a bit, but I think the desire to simplify the data processing process due to easy digestible (if useless) information on the internet has degraded the abilities of younger students to read and comprehend data.  Folks are still capable of repeating snippets information, but the ability to comprehend and synthesize has definitely taken a nosedive over the last decade. This is also correlated with the inability (or unwillingness) of students to read longer texts. Textbooks are rarely read anymore, for example.

Simplifying data presentation in sciences will IMO further contribute to that decline.

 

[Sensei]

Quote

How to read papers

In English? From left to right and from top to bottom..

ps. Almost every day I see a guy who just reads Google snippets created by AI or whatever, and it pisses me off.. RTFM! He can't read something that is more than 20 words long. Born after 2000. Born with ADHD. Glued to a cell phone. If something takes more time than reading Twitter/SMS, they give up..

[CharonY]

2 hours ago, Sensei said:

In English? From left to right and from top to bottom..

ps. Almost every day I see a guy who just reads Google snippets created by AI or whatever, and it pisses me off.. RTFM! He can't read something that is more than 20 words long. Born after 2000. Born with ADHD. Glued to a cell phone. If something takes more time than reading Twitter/SMS, they give up..

That has been the trend for a while. I think the one born around 2000 are roughly at the brink where you still can find folks who area able to at least learn what is needed, even if it seems harder to mentor than the generations before. But looking at the current freshmen crop which is roughly just 5 years later, there is a noticeable drop. I think a part of it is just accelerated by the pandemic. But what I hear from current high-school teachers, the generation after that is going to be a doozy.

Reading is a big issue, as so many other skills are connected to it. But it seems that school boards have been pressuring teachers to drop deadlines for assignments and other things that are considered to be affecting student's mental health. And this is translated into university and those kids will break during actual work.

[Trurl]

9 hours ago, CharonY said:

That is a horrible idea. A paper is a cleaned up synthesis of often a lot of work. Looking at the raw version of it, would be entirely incomprehensible to except for the person generating int. I have to keep telling my students to clean up their notes as often I barely understand what they mean and over time, they will also forget what their notes mean. Research is almost never linear and 80-90% of the material never makes it into a paper for good reasons.

I agree with your reasoning but what if someone writes a math proof then writes in the margin how easy the complete proof is?

Like high school papers when they make you include your notes to prove you did the work. Your notes may seem unnecessary but include them in a hyperlink so you can still refer to them. Imagine if you discovered something big in 100 years your original notes would be gold. To me it is like when comic books first started: they printed the comic book and trashed the original art.

I understand published research is important. But it seems you have to be indoctrinated to the system in grad school.

Any good suggestions on what would be a good article to read?

[CharonY]

55 minutes ago, Trurl said:

I agree with your reasoning but what if someone writes a math proof then writes in the margin how easy the complete proof is?

I don't understand that. I am not that familiar with math papers, but I would assume that any paper with a proof would report on it in a way that can be followed. Why would you put anything in the margins? if it is relevant it should be in the paper. I have no idea why the comparison with high school is supposed to do.

[Trurl]

13 hours ago, CharonY said:

I don't understand that. I am not that familiar with math papers, but I would assume that any paper with a proof would report on it in a way that can be followed. Why would you put anything in the margins? if it is relevant it should be in the paper. I have no idea why the comparison with high school is supposed to do.

I was joking. There is a math paper where the mathematician proved a problem then included another problem that he claimed to prove with ease but no one saw it and this proof is still unsolved.

i will have to look up his name and conjecture. But if someone remembers his name post it here. It is very famous.

[studiot]

Just now, Trurl said:

I was joking. There is a math paper where the mathematician proved a problem then included another problem that he claimed to prove with ease but no one saw it and this proof is still unsolved.

i will have to look up his name and conjecture. But if someone remembers his name post it here. It is very famous.

Are you thinking of Fermat's Last Theorem  ?

 

Wiles proved it in 2003.

 

https://en.wikipedia.org/wiki/Wiles's_proof_of_Fermat's_Last_Theorem

 

But I don't see why that was a joke.

 

Edited Tuesday at 07:18 PM by studiot

[Trurl]

14 hours ago, CharonY said:

I have no idea why the comparison with high school is supposed to do.

Well I was just trying to stress the importance of notes. When I was in high school we turned them in to prove our work and prevent plagiarism. But that wasn’t my point. I was just showing that if the research paper was written as a journal (like a Wikipedia page the writer of the paper could include all data sets and math derivations. You could still have a main paper but link to the parts of the notebook, but the reader could browse the notebook and pull up the research the reader was interested in.

This isn’t my idea. Research papers still have their place. But I think this is how the Internet was supposed to work.

It has historical purposes too. What if you wondered how Einstein thought of the observer outside the train and you wondered how Einstein originated the idea. You look at the scanned notes and Einstein writes, “Today I was reading a good science fiction book. It had a guy riding a beam of light.”

In college we learn to create a journal and keep pictures, drawings. math problem, experiments, and data. I’m not saying not to write a research papers. Instead I am saying to include the entire project’s work. And write the paper for a larger audience. With the storage of today’s computers this is not an issue.

And historically we’d know what made these good discoveries. There once was a mathematician who passed away and his maid thought his desk was too cluttered and burned all his math papers and notes.

Einstein just passed away, let’s erase his chalkboard. And maybe steal his brain.

I understand the tradition research paper is important to share developments and promote science. I just think we should augment it with as much information we can.

[Trurl]

1 hour ago, studiot said:

Are you thinking of Fermat's Last Theorem  ?

Yes. That’s the one.

This is the same book I recommend in a thread of Riemann theorem.

Same book that one mathematician passes away and the maid burns all his papers.

As for Einstein, Walter Isaacson writes about the science fiction book that influenced the General Theory of Relativity.

I think Steven Levy wrote about Einstein’s brain.

And another book that talked about network intelligence and how it applied to large data sets like telescopes pictures. It takes sharing the data (all data) among many minds. Like bioinformatics and protein folding. Unfortunately AI is threatening to take this over, but that is another topic.

[CharonY]

1 hour ago, Trurl said:

was just showing that if the research paper was written as a journal (like a Wikipedia page the writer of the paper could include all data sets and math derivations. You could still have a main paper but link to the parts of the notebook, but the reader could browse the notebook and pull up the research the reader was interested in.

I think you have no idea how many notes are generated in the course of a writing a paper and how much is not included because it would make the whole thing unreadable. There is a bigger point in putting somehow preserving even failed experiments into a repository to ensure academic integrity. But again, considering how much work really is just random thoughts and failed experiments (most of the time due to inexperience) it would be take a huge amount of time to create a dump that no one ever looks into. Folks look at Einstein's note because he is famous and folks are curious about him, not because it is necessary to understand the science.

What I take away form the average student lab book is that they really need to learn to read and write better. Not everything is information and a key element of science is to condense meaningful information. If you want fluff to distract you from information, you got the internet.

[studiot]

Just now, CharonY said:

I think you have no idea how many notes are generated in the course of a writing a paper and how much is not included because it would make the whole thing unreadable.

Hear Hear.

 

I don't know what they do nowadays but when I first went to grammar school back in the early 1960s we were issued with exercise books to write in.

I was suprised to receive two exercise books in at least one subject (definitely History was one).

We were supposed to write 'rough notes and working' in class in one book and put a 'fair copy' of any exercise essay in the other for marking.

I suppose this was in part preparing us for that method CharonY mentions.

[Trurl]

Well I have no experience in publishing papers. I was interested in this thread to learn the trade. I have used APA format and basic labs. My point is that the way papers are written today most of them are only written for such a small audience. If it takes me a week of studying and research to gain an understanding that is why people put it in AI and read a paragraph summary. If this happens in my belief this is going to wreck the whole system.

I know what you guys are saying: a paper refines the information. But I am saying link to support information. If you worked it out on the computer it is probably organized to some extent.

Microsoft was working on a timeline that will screenshot every thing you ever did on the computer. This system does what I am describing. I don’t like that AI will probably organize and train on it. Off topic would it be beneficial for AI to train on failed data and experiments as well as what we put into proper format?

Personally I believe AI is going to fudge up the whole research process. Sure it can do genetics and protein folding, but it does fit your point that we are there to refine the data and can’t read everything. But in my point I didn’t want AI to take our jobs. I wanted to read the paper and see where the writer failed in there experiment failed because sometimes that is important if you have been working on something similar.

[CharonY]

14 minutes ago, Trurl said:

My point is that the way papers are written today most of them are only written for such a small audience.

Define "small". I think it is more accurate to say that it is written for experts and not the general public. No amount of random notes will change that. In fact, it will only confuse folks more as you need knowledge of specific laboratory/group conventions to make sense out of notes. We often use shorthand that is not used much outside of our lab for example. Again, that "supporting" information won't help you. Look, the bottom line is that to papers are for advanced audiences and the only way to understand them fully is to gain at least some level of expertise. There is no simple or convenient way to convey expert level knowledge to a layperson without losing information. If that was possible, we wouldn't need higher education in the first place.

18 minutes ago, Trurl said:

I wanted to read the paper and see where the writer failed in there experiment failed because sometimes that is important if you have been working on something similar.

Look, this is an entirely different discussion and it goes exactly against your point of broader audiences. The folks interested in why certain approaches are a specialized sub-section of an already specialized sub-section. And in 90% of the cases in the wet lab it is because someone messed up during pipetting, or got distracted during the process. Again, if we added each failure as supplementary information you could have a ten-page article with 800 pages of supplements. Add to that discarded raw data you would need a few gigs of downloads for each simple thing.

Now, I will say that there are some journals who try to publish negative outcomes and I do think that there is some value in that. But again, these would be well-vetted experiments and not just a bunch of random outcomes. I have no idea how your thought on AI ties into any of this, though. It is not like that the AI can make sense of these things easier. I suspect you would change your mind if you have actually seen how records are done in the lab.

[studiot]

Just now, Trurl said:

Well I have no experience in publishing papers. I was interested in this thread to learn the trade. I have used APA format and basic labs. My point is that the way papers are written today most of them are only written for such a small audience. If it takes me a week of studying and research to gain an understanding that is why people put it in AI and read a paragraph summary. If this happens in my belief this is going to wreck the whole system.

I know what you guys are saying: a paper refines the information. But I am saying link to support information. If you worked it out on the computer it is probably organized to some extent.

Microsoft was working on a timeline that will screenshot every thing you ever did on the computer. This system does what I am describing. I don’t like that AI will probably organize and train on it. Off topic would it be beneficial for AI to train on failed data and experiments as well as what we put into proper format?

Personally I believe AI is going to fudge up the whole research process. Sure it can do genetics and protein folding, but it does fit your point that we are there to refine the data and can’t read everything. But in my point I didn’t want AI to take our jobs. I wanted to read the paper and see where the writer failed in there experiment failed because sometimes that is important if you have been working on something similar.

This thread is about reading papers.

However I am happy to see it extended to writng them as that is the other side of the coin and the discussion is quite good.

 

I am reading a new book by a particle physicist at CERN  called 12 experiments that changed the world.

 

It starts off with Roengten

Quote

Now he had found something no one else had seen before.

...

He spent seven intense weeks in his lab, ocasionally being reminded to eat by his wife. Apart from these interactions,he was working almost entirely alone, and he remained silent about his research.He didn't tell his assistants, let alone his international colleagues.

...

The only report of him speaking about his work was to a good friend to whom he simply said "I have discovered something interesting but I do not know whether my observations are correct."

 

Many work like this today for fear of being gazumped.

So how exactly is an artificial dumbness going to produce a paper on any such work ?

AI only knows what has been fed into it.

 

I am responsible for a handful of papers, all of which contained original (though of much lower import) work.

Ask AI "How to use the fifth quadrant when converting eastings and northing to bearingas and distance or vice versa ?"

 

 

[Trurl]

1 hour ago, CharonY said:

Define "small". I think it is more accurate to say that it is written for experts and not the general public

I stand corrected. I always thought publishing it to promote the idea. For math I’m not an expert or authority but I know more math than the average person. But what is the purpose of righting a paper only experts understand? I understand writing formally and documenting your work, but it seems counterintuitive. Especially now that print media is struggling. The only reasons I can think of is because it is a trustworthy source and it supports a community even though the community is small because nobody else understands what they are saying😝

I talked about AI because Sensei talked about people relying on it to research. But if the operating system is going to record every screen this will change how we format and share data.

1 hour ago, studiot said:

am reading a new book by a particle physicist at CERN  called 12 experiments that changed the world.

 

It starts off with Roengten

Again by experts for experts. I wouldn’t understand. But this would create a niche writing books to explain to everyone else what the heck they are talking about.😉

1 hour ago, studiot said:

I am responsible for a handful of papers, all of which contained original (though of much lower import) work.

Ask AI "How to use the fifth quadrant when converting eastings and northing to bearingas and distance or vice versa ?"

I would read your papers.

I will plug the question into the prompt. AI isn’t good at math yet. Scientific discussion is a good thing that AI can’t do. But AI will soon be feed all our information. The only advantage we will have is that we still control how the information is used.

But I am biased against AI. It seems like they are ripping us off and a select few are selling it for billions.

But any time you share your idea it can be used anyway the reader wants. But if don’t share it is no fun.

[swansont]

18 minutes ago, Trurl said:

But what is the purpose of righting a paper only experts understand? I understand writing formally and documenting your work, but it seems counterintuitive

You are getting the word out to the audience that can understand and possibly use your work. Work that’s often on the cutting edge.

Quote

Especially now that print media is struggling.

Not really any overlap here with traditional print media. The work is published in specialized journals.

[studiot]

4 hours ago, Trurl said:

Again by experts for experts. I wouldn’t understand. But this would create a niche writing books to explain to everyone else what the heck they are talking about.

Actually no it is not a book for experts. Anyone can gain from it. I often find it fascinating learning the story behind many discoveries.

It is not written in research paper style, but is 'popsci'.

You might be suprised to find what can be learned when competent  scientists put out their thoughts this way, as opposed to journalists and others.

Of course that is not to say that all scientists are 'good' and all journalists are 'bad'.

The great scientific discoveries are often made when someone (not necessarily a scientist) spots something that others have missed.

My point about Roengten was not what he discovered (X Rays) but what he did with the information and how he published his paper, which is a very interesting and human story.

 

4 hours ago, Trurl said:

I would read your papers.

Here is the short fifth quadrant paper. It was published in the Empire Survey Review in 1986 so has yet to reach Big G and M$.

This puplication then had a circulation of perhaps ten thousand surveyors around the world.

You should have no trouble with the maths.

Edited 18 hours ago by studiot

[Trurl]

Nice paper studiot. I had to read through it a couple of times but I think that is just me.

In your flow chart why did you not loop from the end (use theta = 360 - wcb) back to the start (wcb > 90)?

 It takes a little bit of work but for the most part I follow. Definitely a learned talent. I believe I am decent at math with a modest education, but when I try to explain it no one knows what the heck I’m talking about. After Christmas I’m going to get the how to read research paper book you recommend. It is not called how to write research papers because no one can explain how to do that. You have to study examples. 

[CharonY]

17 minutes ago, Trurl said:

It is not called how to write research papers because no one can explain how to do that. You have to study examples. 

What do you mean? There are tons of books on writing research papers. There are general books on the craft of paper writing and specialized ones for each discipline.