[whatshisface]

This is an article about ML research, and the emphasis on branding and marketing your paper wouldn't fly in any of the fields people think of as scientific. Could you imagine someone saying, "be sure that the graphic for the molecule in figure 1 is 3D and has bright colors?"

The most disturbing thing about it is the way advice to forget about science and optimize for the process is mixed with standard tips for good communication. It shows that the community is so far gone that they don't see the difference.

If anyone needs a point of reference, just look at an algorithms and data structures journal to see what life is like with a typical rather than extreme level of problems.

[0xfffafaCrash]

Strongly disagree here. While I haven’t published e.g. particle physics work, I have authored/coauthored a number of peer-reviewed papers in other topics generally considered hard science (and published in ”high impact factor” scientific journals). This article and series is just as accurate about how ”Science 2” works outside of ML in my experiences. Branding and marketing is a very major factor in everything from grant funding to research publishing in academia.

[whatshisface]

There's a used car sales line somewhere and you just have to be careful to not cross it. Yes, there are rewards to good communication but if it becomes the sole purpose (communicating "read me") that's too far.

[chairhairair]

> Could you imagine someone saying, "be sure that the graphic for the molecule in figure 1 is 3D and has bright colors?"

Chemists are extremely brand-aware regarding their figures.

In synthetic chemistry many chemists could guess the author based just on the color scheme of the paper's figures.

For instance, look at the consistency here: https://macmillan.princeton.edu/publications/

And it comes with rewards! The above lab is synonymous with several popular techniques (one, organocatalysis, which garnered a Nobel prize) - the association would be much less strong if the lab hadn't kept a consistent brand over so many years.

[mbforbes]

Oh my, those figures are gorgeous! Thank you for sharing.

[rrr_oh_man]

> This is an article about ML research, and the emphasis on branding and marketing your paper wouldn't fly in any of the fields people think of as scientific

The number of accepted papers is absolutely currency and measure of worth in academia.

[Fomite]

Interestingly, one of the pieces of advice, about having a punchy title, is a double edged sword. There's some data suggesting papers with "clever" titles have an easier time getting published, but accumulate fewer lifetime citations.

Both of which are currency.

[eru]

I suspect there's a selection bias here.

Silly example: if I ever find out a prove saying that "P=NP", that will also be the title of my paper. No cleverness required to grab attention.

If I have a more pedestrian result, I'll think up some clever title.

[spookie]

As someone doing a literature review I can safely concur with this. Fancy titles often do not include the most obvious terms of what you are searching for, leading to fewer results from your query.

[karmakurtisaani]

The reason could be that clever titles add "novelty", but not much substance. Publishable, but not citable.

[Majromax]

> The reason could be that clever titles add "novelty", but not much substance.

Another reason might be that clever titles stand out as bold claims, working counter to the common practice of academic humility. If a paper seems to be downplaying its own significance, then why should a casual reader (or reviewer, at first impression) give it the benefit of the doubt?

That's not to say that papers should over-claim, and I suspect that doing so might lead to a harsh counter-reaction from reviewers who feel like they've been set up to have their time wasted. Nonetheless, "project confidence" might be good practice in academia as well as one's social life.

[whatshisface]

That's true as a zeroth-order approximation, but even sticking to trivially quantified values your citation count is more important (that's maybe first-order), and on the level of your reputation the question you need to ask is, "will people feel like my work actually benefits them?"

[pazimzadeh]

It would be great if he’d shared the actual reviewer comments.

> Could you imagine someone saying, "be sure that the graphic for the molecule in figure 1 is 3D and has bright colors?"

I doubt the reviewers asked for that, but yes that kind of thing happens all the time prior to publishing and there’s nothing wrong with it. If it reduces the amount of time it takes to understand the paper then do it..

[Fomite]

I have definitely said that people can't resist a network diagram, and "people love a good map", and I'm not in ML research. There are things that appeal to people.

This tends to not manifest as "We need one of these" but "If we have one of these, lets be sure to use it."

[whatshisface]

People like witty epigraphs underneath chapter titles too, and that's great. Now imagine saying, "the difference between this paper getting accepted or rejected is the presence or absence of a network diagram..."

[intended]

The difference between getting fatigued reading one paper vs one that makes it easier to see the point you are making.

Evaluators are human.

[auggierose]

Yes. That is indeed the problem, and might be resolved sooner than later.

[Fomite]

I mean, at one point I was presenting some research next to someone with a network diagram, a map, and a phylogenetic tree and my comment was "That's going to win best poster" and I was right.

[chneu]

my spouse works in academia and publishes regularly.

This article is spot on. what are you talking about? have you ever published a research paper and gone through peer review?

[whatshisface]

I think a lot of people are reading the presentation advice and thinking "yeah, I work hard for good presentation too," without realizing that the reason content hasn't even been mentioned is that the author really is describing ML accurately.

I think it's ultimately due to a lack of theory, which creates the expectation that the results from trying an idea will be a random draw. From that point, you get the behaviors of trying as much as possible and taking each attempt as a fixed object to then go try and get over the threshold.