Adam Jones|HomeBlog

Reflections on my 7-day writing challenge

Headshot of Adam Jones

Adam Jones

I set myself a challenge to write 7 blog articles in 7 days, and promised to write a follow up reflecting on the experience. This is that follow-up.

I did it!

I not only met, but exceeded my target. I wrote 10 articles, although I only published 8 within the 7-day period. I'm proud of the articles and believe they mostly hold up to scrutiny against the criteria I set for myself:

  • Honesty: saying things that I fully believe in, and being explicit when I'm not sure about something.
  • Verifiability: making it easy for readers to check my claims.
  • Clarity: using simple language and consistent terms, and linking to relevant further reading.
  • Importance: writing about topics that could matter to someone, with some element of novelty.

Here's a breakdown of how I feel each article performed against these criteria:

Article TitleHonestyVerifiabilityClarityImportance
Why having a human-in-the-loop doesn't solve everything✅✅✅✅
Why AI systems should get more things wrong✅✅✅✅
Asking for help✅✅✅🤷1
Avoiding unhelpful work as a new AI governance researcher✅🤷2✅✅
Government departments should say they don't care✅✅✅✅
Proof of address is nonsense✅✅✅✅
How to fix proof of address✅✅✅✅
AI as a corporation (or, an intro to AI safety?)✅✅✅✅
Addressing digital harms: a right of appeal is not sufficient✅✅✅✅
Diagnosing infectious diseases with CRISPR: SHERLOCK and DETECTR explained✅✅✅✅

Things that went well

Key actions that made this go well were:

Reserving time: A week off work was necessary for me to dedicate so much time to writing. Now I'm back to work I know I'll write less, but I can still carve out time for writing. I also think I'll use this time better: I've gotten faster at writing through the additonal practice.

Maintaining an idea backlog: Having an idea backlog of things I was genuinely very excited about was incredibly useful. It gave me a wealth of topics to choose from each day (I initially tried planning which articles to write on which days, but then it felt more like a chore: so I switched to just writing whatever I was most excited about on the list on any given day). I didn’t manage to change the length of the backlog much: I ended up adding new ideas at about the same pace as I wrote things.

Setting targets: SMART goals work! Having a target for this challenge was quite motivating. Aiming for a more achievable number felt more motivating for me than setting an overly ambitious target: something in my brain quite likes thinking the job is easy, and being able to exceed targets.3 Also, focusing the target solely on quantity worked well, given I knew I’d find it hard publishing something I didn’t think was high-quality.

Things I’d like to do better

If I did this challenge again, I don’t think I’d change how I approached it. However, I could make future writing more impactful by:

Promoting my writing: If a blog is posted on a website, and no one is around to read it, does it make an impact? Probably not. Sharing my work on Hacker News and LessWrong has previously driven a lot of traffic to my blog, and seems useful for broad outreach. The inbound links are also likely to result in better SEO (and also being the only person in the world who writes on certain topics helps with this a lot!). One difficulty is that there’s not always an obvious place to share a lot of my kind of content.

Taking action myself: There's a risk that I write things and consider the job done. However, the real work often comes after the writing. For example, for my article on improving proof of address, I still need to find and contact the relevant people who could change the rules.

Practising plain English writing: I want to get good at writing in plain English. I’ve been thinking about this because I’m working on improving the project phases of BlueDot Impact’s courses, where I think participants would benefit from better communicating what they’ve done. But I realise I have a lot of room to improve here too!

Improving on-site discoverability: I've published 33 articles, but they’re all just presented in a list on my site. There’s no easy way to filter or search content, and people new to my blog don’t have an obvious place to start. I think it would be worth adding a 'start here' section at the top, and potentially a search feature.

Footnotes

  1. It’s arguable whether I met this criteria. I think this is certainly a useful article for future me, and I have already sent some people to it - but it’s not really for a decision that is ‘important’ and is instead just a time-saving device for myself. ↩

  2. Many claims in this article were not directly verifiable. However, I think the reasoning was transparent, so that people can scrutinise these claims a bit better. ↩

  3. Which often works great, but can also materialise in negative ways, like being a perfectionist or overworking. Half-assing it with everything you've got is a good article to remind yourself that you don’t always want to aim for maximum quality. ↩