Hacker News new | past | comments | ask | show | jobs | submit login
It's Time to Start Writing (2019) (alexnixon.github.io)
280 points by dhotson on July 21, 2020 | hide | past | favorite | 57 comments

I've given significant thought to the writing problem, and I firmly believe the lack of writing is dramatically slowing the pace of innovation in the world. Writing is really Amazon's superpower. It's why they get big things done quickly.

Both the author and the audience greatly benefit from the writing. Writing clarifies and structures thinking – helping the reader understand the points the author is making.

As a bit more context, I used to work at McKinsey and much of my job fit into two roles: (1) translating what employees were thinking into something executives could understand, and (2) making PowerPoint slides. In other words, I was often there because employees couldn't write well. But, I also found PowerPoint lacking – it's hard to get some of the more important points across (creating some confusion) because it doesn't allow for longer-form thought.

I've put a bunch of thoughts together on why writing is important and how we fix our education system to make people better writers. It's based on my experiences supporting tens of thousands of students on improving writing skills – https://bradsblog.com/2020/05/15/1-writing-is-the-most-impor....

Amazon has a few superpowers. One is writing, but the other is so many metrics. I worked at AWS and saw how the weekly metrics meeting was ran, with execs questioning why some metric, out of 10s of thousands, was behaving in a certain way. A good product manager had a good reason, an acceptable one found a reason quickly after the meeting. Without an understanding of the business writing is worthless, and one of the ways Amazon creates this understanding is comprehensive metrics gathering.

It is the combination of business understanding and writing that is Amazon's superpower.

Agreed that understanding the business is critical. Writing is one of the best ways for someone to develop a better understanding of the business - e.g., what metric are we tracking and why?

Overall, I'm not saying that there aren't a bunch of other important things to running a business, just that great writing makes running a business far easier.

On a side note – implementing performance management gets crazy gains without requiring process changes. I used to get an immediate 30%+ productivity bump when I'd implement performance management (e.g., metrics, daily huddles) in a place that had little of it.

Hey, just want to say I read your blog posts and was searching for a call to action, how do I sign up for Prompt?! I kept asking myself while reading the blog posts. Then I tried googling Prompt / Prompt writing, that name is -not- favorable to SEO. When I hit the landing page, I found out that Prompt is meant for students, not people already in the workplace. I don't even know how it would work, but I was hoping there was some way a working professional could improve their writing. That said, when you do hit the Prompt landing page, the I am a student is highlighted and there is no other call to action or thing to do, I guess you are supposed to sign up in the top right? Anyways, wish you well!

Hey! Thanks for the comment and going to the site (Prompt.com). We're focused on students for now, but we will be launching supports for professionals in the future. We're almost all B2B sales, which is why you found it harder to find us than you'd have probably liked.

As it turns out, most people aren't similar to the Hacker News population – they won't seek out and act on feedback on their own. As such, we've found the best way to improve writing skills is to require people to get feedback and act on it. This means working with K-12 and higher ed institutions.

Are there any great examples of how Amazon writes their white papers? I've heard of the Press Release method, but are there any examples online? Also curious if there are white papers that don't fit the Press Release model.

Here's an AWS article explaining how root cause analysis is done at Amazon: https://aws.amazon.com/elasticsearch-service/resources/artic...

> The technique consists of asking the question “Why?” iteratively until you get to the root of the problem. Let’s see a quick example:

Problem: The website is showing error 500.

1. Why? Because the web framework’s routing component malfunctioned.

2. Why? Because it requires another component, which itself malfunctioned.

3. Why? Because this component of the web framework requires the intl extension, which isn’t working.

4. Why? Because it was accidentally deactivated after the server software got updated.

worth noting that there are reasonable criticisms of the 5 why's method. complex system failures often do not have a single root cause.

What would your recommendation for a professional who wishes to improve his writing? Word can mostly fix my grammar.

The philosophy of Deliberate Practice applies very well to writing. There are some online writing courses you can take (e.g., David Perrell has one), but there's a lot you can do on your own (and with the help of some friends). Here are the two biggest things.

1. Start with an outline. This will help you cement the most important points of your writing/argument in your mind and enable you to start filling in the gaps. Often, I'll spend a lot of time getting the major points I'm making very crisp before I write a full draft. Doing this will also help you identify what's missing. The most important parts to focus on are (1) where a reader may disagree (i.e. where you'll need to make your argument stronger and back with evidence), and (2) where your reader will have questions that you'll need to answer.

2. Get feedback and revise 2-3 times. Getting feedback and acting on it is the single most valuable thing you can do; however, most people don't know how to provide writing feedback. They'll focus on grammar unless you specify what you want the reader to think about. There are two strategies here – (1) Provide your reader with a list of questions you want them to answer. I always use the following four: What did you learn? Is what you learned compelling (if not, why wasn't it compelling)? What didn't you learn that you wanted to learn? Where was it clear/not clear? (2) add specific comments/questions within your writing in places you are less comfortable with and want feedback.

Thank you for the suggestions. What would be a good strategy for finding people to give feedback early?

I'd test out some of your more thoughtful and detail-oriented friends. You may find one or two who you feel are adding value and pushing your work forward. If you're going to pay for someone, it's very difficult to find a good editor. Most "editors" are only used to providing sentence-level feedback and may not have enough context about the subject your writing about to be helpful with content feedback. At Prompt, we find only 2% of the thousands of Writing Coach applicants we receive are good enough to make it into our network – i.e., the writing feedback skill is really hard to find.

The sure advantage of a powerpoint is that it gives much less way to criticism.

Imagine someone wants details on a graph or table, the presenter replies orally and can often remain as vague as desired, possibly making the requester look like an a*e if he/she insists.

How do you know? Does Amazon have a blog? Where do you read their documents? Aren't they internal?

There's not much out there on the specifics of Amazon's approach to writing, but there are bits and pieces around.

Jeff Bezos's 2017 shareholder letter highlights some of his thinking around 6-page narratives: https://www.sec.gov/Archives/edgar/data/1018724/000119312518...

This Slab article by Ben Bashaw is pretty good on the topic: https://slab.com/blog/jeff-bezos-writing-management-strategy...

Scott does a good job of some of the major points of Amazon writing based on his time there (this was written based on a previous HN discussion on Amazon writing): https://blog.usejournal.com/writing-docs-at-amazon-e02580861...

This aligns well with a lecture from the University of Chicago I watched recently[0], which explores how many academics use writing as a tool for thinking (What the article describes) and don’t think about the reader. Writing is an incredibly powerful tool to form thoughts.

However, as the lecturer in the video details, don’t expect the writing done for thinking to be useful to readers! When we think, there is usually a certain incrementalism, whereas when we read we’re trying to resolve dissonances in a current mental model. That’s a key distinction, it means you need to make people care or understand why their mental model is broken when writing to informs reader. When we write for our own thinking, we usually understand the general problem already and are just traversing the problem space to better understand different components.

[0] https://m.youtube.com/watch?v=vtIzMaLkCaM

> "many academics use writing as a tool for thinking... and don’t think about the reader"

That would explain an awful lot about my university reading experience.

Don't believe me, watch the talk! If this resonates with you, the talk will both resonate and leave you better equipped to fix the problem moving forward. His role is to help academics write in ways readers want to read.

Interesting. I was taught to consider the audience in communication, but I think I fail to consider their mental model most of the time.

You write once for yourself/thinking, let it sit overnight. Come back the next day, and without looking at the original text, write it again for the audience.

This video alone is worthy of its own topic on Hacker News.

Agreed, this was great.

Thanks (To both of you)! I've submitted [0] it and hopefully others resonate with it as well and that fosters more discussion around it

[0] https://news.ycombinator.com/item?id=23917587

One thing that often gets missed with PowerPoint is that it’s a fundamentally different mode of communication than writing, but people don’t treat it that way. One of the reasons it’s so hated is because people are trained to write essays and emails, and they try to build ppts in the same way.

Done well, slides can be incredibly informative (link below). Because it’s visual, you can present the shape of an argument well before having to articulate the details, a tremendously powerful tool for communicating complex ideas. Because it’s nonlinear, you can include a level of detail that would be tedious in a document.

Yes, these benefits often enable people to spout half-understood BS. But the medium itself is sticky for good reason - it’s powerful.

Edit: link https://medium.com/the-mission/the-greatest-sales-deck-ive-e...

People spend far more time learning to put together "reports" in powerpoint. A slide deck is a sales tool. If you're selling a product, sure, make a deck. If you're communicating any nuance or subtlety, you will be hampered if you attempt to make it out of powerpoint.

Bad powerpoints lead to bad decisions, and Edward Tufte addressed this years ago. https://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=...

You could probably find an even worse ppt that justified the Iraq invasion, but nobody thinks they're dumber than the DoD. What about NASA though? Incredibly smart, hardworking, technically savvy engineers saw their recommendations get ground to dust by powerpoint. They couldn't write a good technical powerpoint. You can't either. Write the whitepaper instead. Use powerpoint for pretty pictures and graphs.

> What about NASA though? Incredibly smart, hardworking, technically savvy engineers saw their recommendations get ground to dust by powerpoint.

The problem this didn’t address though is when in-depth white papers are often written, they may not be read (or only read in a cursory fashion) because the audience may be stretched too thin on time.

PowerPoints abbreviated style is both a strength and weakness.

If making the document faster to read is all it took, we'd see 140 character technical reports.

The difficult thing is understanding. Powerpoint masks this, so readers don't go looking for extra time to comprehend. Powerpoint also encourages poor communication (eg auto-resizing text blocks as you add more/longer entries) so the writer doesn't think deeply about the document they're making. It's a strength in some fields, and a weakness in others.

Your boss asks for a deck? Sure, make the best damn deck you can. We've all got to earn a living. But one day, when you get to set the meeting agenda, be better to your reports than your boss was to you.

>If making the document faster to read is all it took, we'd see 140 character technical reports.

I suspect we’d then see the same hate talking about how all these executive summaries hamper understanding by glossing over nuanced details.

Granted, PowerPoint is often misused. But just because I try to hunt with a shotgun at 300 meters doesn’t make the shotgun a piece of shyte. I just misapplied the tool.

I can say, if you have a busy boss and constantly give them heaps of white papers as your communication tool, it’s a good way to ensure your concerns are rarely heard let alone addressed.

I think there’s a natural dichotomy between limited bandwidth and understanding complex topics. I personally don’t think white papers solve that problem.

White papers don't have to be 10 pages each. Some things are two pagers, some need 5, and some really beefy topics need more.

The problem is that in addition to the clay thrower, the shotgun comes equipped with a scope and a 300 meter paper target. Yes people are fools for misapplying the tool, but the manufacturer is to blame for misleading their customers to the tool's capabilities.

The problem is that bad writing is immediately obvious to fluent speakers, but bad powerpoints are not. Powerpoints are also susceptible to institutional styling (look up the face-melters that the American DoD designs).

I would wager that a busy boss will pay more attention to your ideas if they're presented as a tight page or two. Good writing is hard to do, and takes time for almost everyone. Presenting topics in a way that makes their complexity evident helps people allocate their attention correctly.

White papers don't solve the problem, but at least they don't make it worse.

If I had to guess, I think the difference in the way you and I are looking at this issue is that I do not think Powerpoint should be used to replace reports, as they both have a distinct purpose. Rather than display report-like information, I think their best use is to supplement discussion, where the discussion is intended to portray more nuanced information.

We're really of the same mind here, but the problem is that people read "powerpoint is best used to supplement discussion" and think "I'm using powerpoint in a discussion, so I'm ok"

I should have probably started with a disclaimer: I'm a PhD scientist at a major American R1 university. I collaborate with researchers and business partners around the world. I regularly write/share powerpoint decks, because it's the right tool for the job when I have figures and pictures of my work to share. When I post design documents, calculations, etc, I always write long-form, typically using the Tufte-handout class.

>The problem is that bad writing is immediately obvious to fluent speakers, but bad powerpoints are not.

This is a very good point

Totally agree. In some cases, explanation technically is much more effective when using PowerPoint or any kind of visual presentation rather than a wall of text.

100% agree

I started writing investment thoughts last year, and the more I write, the more I realize it is a superpower.

It creates clarity of thought, accountability, and enforces longterm thinking.

AND, if you do it publicly, it can lead to unexpected conversations & opportunities.

Link below in case anyone's interested - always appreciate any feedback :) https://playingfordoubles.substack.com/

I like what you've done, so don't take this the wrong way. I'm playing devils advocate here because I think it helps strengthen our understanding.

One thing I've noticed is that 1) your stock picks are almost exclusively in the tech sector and 2) your initial pick was in April in 2019 so there isn't a lot of data to drawn conclusions from.

As an example, simply buying XLK (technology ETF) on the same start date gives an absolute return of roughly 43% and gives potentially less systemic risk (I haven't calculated the exact risk numbers for your portfolio, but it's probably a safe bet given it's diversification).

Instead of measuring strictly on % return, I would suggest measuring performance in a way that factors in volatility as well.

After a cursory look, it would be helpful to present your data with a risk adjusted metric as well as providing some longer backtesting

This is great. I've written previously about why I write as a researcher [1], but none of my reasons were this author's reason: that narrative structure clarifies understanding. This resonates with me because I've noticed that I remember blog posts better, that I understand the topic deeper, if I really flesh it out with narrative writing: why this model, who developed it, what are the alternatives, etc. It often feels like a waste of time initially, but I almost always find that the process makes me realize there are details I initially missed.

[1] https://news.ycombinator.com/item?id=22033792

There are different kinds of narratives. What you're describing isn't really narrative, it's active critical investigation, which is a different process. It is indeed very useful for problem solving - not least because you can read your notes a year later, remind yourself see why you made specific decisions, and see options you considered and eliminated.

But there's also rhetorical narrative, which is used for persuasion, not clarification. Create a story with a good person and a bad person, or at least good actions and bad actions, append a heavyweight emotional hit of some kind, keep the language as simple as possible, and you'll have no trouble selling your point to at least some people - no matter how nonsensical it is to others.

This is how advertising, political spin, social media influence campaigns, and troll farms work.

It's also how thought leaders try to work, but luckily not many are experts in rhetoric.

It's incredibly powerful because the human brain uses narrative as a kind of alias for information transfer. As soon as an idea is packaged as a narrative - a story, parable, something about people - it immediately becomes much more persuasive than a plain statement of fact.

If you look at company culture it's almost invariably based on a narrative about the goals of the company and the kinds of people that fit in. Some founders deliberately engineer an appealing narrative and use it as a cover story for their own personal interests.

Point being not all narratives are benign, and story telling - not just writing - really can be a superpower. But that doesn't mean it's inherently positive.

Writing helps me to "codify" things.

I've been developing software so long, that my process has become "instinctual." I actually have a hard time explaining it to people on demand.

So I started to write about it[0].

Turns out, there's a heck of a lot of detail in my personal process, and, when I write it out, that detail (and the structure), reveals itself.

Basically, I don't particularly care whether or not anyone actually reads what I write, but the act of writing has helped me to "name" my structure.

Once I can find its name, I control it.

[0] https://medium.com/chrismarshallny

This is why you see design doc culture at Google as well. At my startup I pushed for an RFC culture to flesh out ideas and break stalemates just as the author describes. I called them RFCs because we don't have a tool that versions documents, so if we need to make amendments we mark older RFCs with newer RFC extensions or supersede them.

There is also a large benefit to other engineers getting up to date on a system, why things were done a specific way, why a technology was chosen, what tradeoffs were considered and why.

Could you elaborate on what an "RFC Culture" means to you in a broader sense? I feel like there's a train of thought here that I'm missing and am keen on further elucidation.

What would a documentation or comment-chain ideally be constituted of? How would you change existing docs in a practical way which would reflect both deprecated and replacement functionality?

When you originally considered "RFC Culture" what system did you have in mind?

You do not have a git repository?

The author is proposing writing as a tool for self-improvement and to find clarity.

Sure enough, that will help to find clarity for sure. Just don't expect that it will help to find clarity for anyone else beside the author.

In a fast moving team setting, what most people hate the most is reading. If the context is not clearly set why reading something is utterly important, most people won't even bother to open the page, document, note or whatever.

The mental load is immense these days, and reading adds an enormous plus to that in a very negative way, so most people tries to escape this cognitive load.

Of course, you can use your authority to force people to read, but people will simply hate the experience even more.

This is true for most organizations. Note though that the practice the author describes at Amazon is that everyone gets the document to read at the beginning of the meeting. The structure of the meeting means that (1) someone is forced to produce the narrative ahead of time and (2) everyone has to read it alone together.

That's an odd setup, but it's probably what makes this system work -- an organizational admission that this meeting setup is actually better than walking people through slides. One advantage the author doesn't mention is that this "alone together" mode gives each participant more time to think before having to fit their thoughts into a group discussion, which likely promotes divergent thinking and gives people room to ask important, difficult questions.

I believe your frame of reference is a small environment.

If all the stakeholders fit in the "2 pizza" team, regularly talk/meet, and regularly review progress/ideas/etc. then sure, writing is irrelevant.

Frankly, for me, reading is a far LOWER mental load than listening to someone talk.

Talking is linear, and often stream of consciousness. I can't jump forwards to see if they are drawing the conclusion they seem to be aiming at, and if so, scan the rest for any variations.

I can't jump back, without disrupting others, when they have made a cognitive leap that I didn't follow, to re-read to see if I missed anything.

Often people aren't able to quickly order their thoughts, speak to these points, and stay on point/relevant in the face of q&a when it's a deep technical topic. I'd much rather read their perspective.

That "fast moving" more often than not describes only the velocity, but not the displacement. What's good in moving rapidly in random directions and not going anywhere? Maybe spending some time on quality writing and reading would pay off ten-fold.

I recently read this book, recommended in another of the infinite posts about writing well, and it is truly good:


The chapters from 1 through Cohesion II are better at teaching clarity of thought in writing than any I've read anywhere. Most books focus on grammar and style and ignore the good stuff in between.

For those who like Williams, he also wrote books on how to build arguments and conduct research!

It's one of the reasons I write reviews on goodreads of everything I read. That it may benefit someone else is a secondary consideration. It's mostly for me to think about what I've just read and also something to look back on in the future when I want to see what I thought of a particular book. I always hated English as a subject in school, so I use this for practice.

Totally agree.

On a slight tangent, my writing has improved tremendously since working at Amazon. I’m far from great now, but I’ve always been contrarian and saw most writing as needless fluff. Over time I’ve come to appreciate how writing forces clarity, both for the writer and the reader.

> Writing is thinking. To write well is to think clearly.

This idea of writing as a proxy for thinking is the essence of many misconceptions.

I know many 1st-class thinkers who have such hard time in writing.

What separates most people from good writing has very little to do with style, grammar, local sentences structure, word selection, or even content per se.

Most people can't write well because they don't know how to control the logical sequence in which they present their ideas.

And that is the single most important act necessary to clear writing.

I shared more on what that means here in a recent essay [0].

[0] https://pulseasync.com/operators/share-written-ideas/

I think the document culture does really force clarity in ideas. I think the lament on slides is often conflated with poorly crafted slides though.

Well crafted slides force a similar degree of clarity when approached with the same level of intention. The problem is that most people are poor presenters, and presenting is a learned skill.

Documents gone bad encourage people to write more words than necessary which, when used for simple tasks, result in large amounts of time wasted preparing something that no one reads.

I think one downside of a document culture is it leaves less room for selling an idea- which may be less important in the decision to light up a new business unit, but put at a disadvantage work that is primarily visual in form.

imo slides are much more effective if you're a good presenter AND you are recording your presentation.

But most people are not good presenters, and so their slides are either a wall of text OR they're an absence of text that make the slides impossible to parse without a recording of the speaker.

So it's not necessarily that writing is the BEST format, it's just not the worse when it comes to the lowest common denominator.

The other upside that the post fails to notice is that it forces everyone to get used to writing, which will lead to better documentation overall.

As a big fan of bullet-lists, for concisely summarizing key points, I'm wrestling with this: "If someone builds a list of bullet points in word, that would be just as bad as powerpoint."

Is Bezos referring to cases where the "why" is not clear, and not explained?

Lots of times I'll use a bullet list, with indented subitem bullets explaining why, or going into detail. I find this format much easier to read than a wall of text. And it makes the hierarchy clear.

Note that the article uses 3 bullet-lists, which were easy to read/scan quickly.

Bullet points are fine as a summary, however there is often a reductive quality to them that gets in the way of understanding. In many cases where you both the responsibility, the authority and the trust of management this may be sufficient. However if what you are saying could be controversial—especially in a large-company setting where you won't necessarily have much face time with all the people who might have opinions—there is no substitute for a well-crafted narrative to bring people along.

I feel like it's important to emphasize the importance of writing as a way to organize your own thoughts and create your own mental maps. This may not be the best way to disseminate information to a team, or share with the world.

I wholeheartedly agree with this premise, but without some guardrails I feel like this advice leads to dumpster fires like Medium.

When I am designing something, I often find myself making bulleted lists when I am unsure of what to do next. I find that my more insightful sessions result in written paragraphs rather than lists. So I agree with Jeff here.

Bezos's comments here seem to echo things that Edward Tufte has said.

I do think presentations ('powerpoints', though Powerpoint itself is pretty horrible) have their place, if done well — but they very rarely are.

Applications are open for YC Winter 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact