Mustafa's Newsletter

By Mustafa Sultan

Creating the Meta Layer

#35・ Nov 25, 2020

issues

Creating the Meta Layer

By Mustafa Sultan • Issue #35 • View online

Friends,

I thought I’d update you on explainthispaper, an idea I’ve spoken to you about before.

A Refresher

Do you keep up with medical literature, or are you like me — and tell people that you do but cannot remember the last paper you read?

Dr Eric Topol told me that reading papers was the single most important habit for his career success.

So me, Stu and Chris are trying to make medical papers easy reads—

Genius, but for medical papers

The Meta Layer 🧠

But there’s more—

Making easy-to-read summaries of papers is cool. Some journals (and grants) already require you to write a layperson summary of your research. Not particularly revolutionary.

Where I really see potential?

The meta layer we’re working on. If you’re not active on academic Twitter, you may not be privy to all the discussion that happens around papers.

Here’s a Twitter thread from a high profile medical AI researcher critiquing a recent paper—

Luke Oakden-Rayner

@DrLukeOR

The authors of the letter spend about 2/3rds of it complaining that they can't reproduce *the model*.

Even when they talk about data access (their only other argument), they are mostly focused on the data "the models were derived from".

Why is this So. damn. weird?

3/12 https://t.co/tNmnXFNraa

11:35 AM - 16 Oct 2020

This is really cool and completely unfindable unless you happen to follow the right accounts.

What You're Missing

So broadly, what are you missing when you open a PDF from a journal website? (And what can our meta layer offer?).

Understanding: Why is this so dry/unintelligible!?
Context: Where does this study fit into the bigger picture?
Education: What do all these terms mean!?
Discussion: What do other experts in the field think of this?
Hidden insights: Is there any interesting behind-the-scenes information?
Next Steps: This is cool. Where can I go/what can I read next?

The Hero: The Meta Layer

How we're applying the meta layer

So it starts simply. You can click the green highlighted text and see our annotations on papers.

This is very cool. But what we’re working on next is even cooler—

The Hero #2: The Author

We’re getting authors to annotate their own papers. It’s not immediately obvious why this is interesting… I mean… they wrote the paper after all. Why do we need more space for them?

Consider this paper from DeepMind—

Clinically applicable deep learning for diagnosis and referral in retinal disease

From interviewing one of the authors, I know that they actually started with a database of 1.3M scans before cutting down to just 14,884 scans.

Why? (I thought big data was 🔥??)

‘Big data’ gets a lot of hype — but what’s more important is dataset quality. 14,884 high quality/relevant images are far more useful than 1.3M unsorted ones.

^ This is an example of the behind the scenes/educational content which is missing from papers — and something we’re bringing to our meta layer.

The Hero #3: Debate and Discussion

Currently, if you take issue with a paper you can a) write a letter (which might be published in a couple of months) or b) vent on Twitter (see previous Tweet).

In the future, I see academics coming to explainthispaper to discuss/critique new papers. As anyone on Reddit knows — you never read the linked article — you come for the comments.

This, I hope, will be petrol on the fire.

So What Next?

We’re working on making explainthispaper as beautiful and curated as Blinkist, with all the community of academic Twitter and the UI of Genius.

Blinkist

More importantly, we need prominent authors to buy into our vision. We’re going top-down, because wherever the authors go — the community will follow.

That means lots of emails, lots of “hey you know X, can you intro me?” and lots of 6AM #hustle.

Other Cool Stuff

🚀 Reflecting on My Failure to Build a Billion-Dollar Company

I enjoyed this article from Sahil Lavingia (founder of GumRoad):

Instead of pretending to be some sort of product visionary, trying to build a billion-dollar company, I’m just focused on making Gumroad better and better for our existing creators. Because they are the ones that have kept us alive.

Podcast

#038 Ferraris to Fitness — Dr Yusef Smith (Propane Fitness)

‎Why switch from Investment Banking to Medicine?

Dr Yusef Smith is a doctor, competitive gymnast and powerlifter. In a previous life, he was an investment banker before packing it in and training in Medicine. Whilst at Medical School he started Propane Fitness — the passive income from which has now…

podcasts.apple.com • Share

All the best,

Musty

Did you enjoy this issue?

By Mustafa Sultan

What I'm reading about.

Tweet

Share

In order to unsubscribe, click here.

If you were forwarded this newsletter and you like it, you can subscribe here.