Weeknotes 368
This week I did:
Learning & development
Our team had it’s first learning and development day this week. I wanted to learn more about AI in the charity sector. I read a few articles going back to 2018, which showed how we talk about AI has changed from being innovative experiments to mundane, and of course I used ChatGPT to summarise them for me. I started a Microsoft Learn course to understand how Azure AI could be used. And I experimented with the three use cases for LLMs; summarising, first drafts, and interrogating data.
Related, I’ve also been working how and whether we should prevent AI bots from crawling our website and using the information as training data. There are two, almost opposite use cases. Pages that we do want to be included in training data, and may even want to think about how we optimise for it. And pages that we don’t want to be included because they contain information about real people that LLMs could misinterpret.
Content discovery
One of the things I (used to) use social media for is new content discovery. As social media is collapsing and the future of large sites seems uncertain, I’ve been looking for a way to decouple that part into a more reliable system.
I set up a Slack workspace, created channels for each of the topics I’m interested in (agile, charity, product management, etc.) and used Slack’s RSS app to post messages whenever there’s a post on a website or medium of the people I follow on Twitter. It’s a shame not every website and newsletter platform provides an RSS feed, but it makes reading stuff from those that do more reliable as I’m not dependent on an algorithm hiding stuff I might want to see.
Next, I’ll think about a way of decouple conversation away from social media.
Productivity
I completed 47 tasks this week across 13 projects, averaging 9.4 tasks a day. The two projects I did the most on each had 7 tasks completed and the least busy project only had 1 task. As I get more into visualising my work in this way I hope it can help me understand how much work I have in progress, not just a vague list of things, but an empirical understanding built on the actual work I’ve done.
I read/watched:
The founder
I like these kinds of films. Origin stories for companies like Apple, Spotify and in this case, McDonalds. The two lessons I took from this; the value stream is bigger than you think it is and where the most value is isn’t always obvious, and that the ‘McDonaldisation’ of products and services in an attempt to make them all the same is only a good idea it’s the fifties and you are McDonalds. Every other organisation should be aiming for variability.
Harmful design in digital markets
How Online Choice Architecture practices can undermine consumer choice and control over personal information by the ICO, CMA and DRCF, goes into how deceptive patterns are used on websites and contravene data protection regulations.
Speaking a second language
Helen Jeffries shared this article she wrote about experience of autism and struggles with communicating from a couple of years ago. One day I’d like to write more about how my autism effects my communication. For neurotypicals, conjunctions like ‘and’ and ‘or’ seem like they are just words they throw into sentences without thinking about it, but for me that are logic gates that explain the path of the point I’m trying to make. So, to answer a question with anything more than a few sentences, I have to map the logic in my head and follow it as I speak. If I loose where I am in my map, it’s really hard to keep speaking. If it was acceptable, I’d use ‘nor’ and ’nand’ in conversation too.
I thought about:
Outputs into outcomes
I’ve been thinking about how to reverse engineer outputs into outcomes. If I can crack it, and create a system for it, then rather than always having to stop people when they start with what they want to do and ask them what they are trying to achieve, I can group up the ‘whats’ into ‘whys’. Then it’ll be easier to check with them if that’s really want they’re trying to achieve.
Changing users
Something occurred to me about how we talk about users is a bit wrong. I think I’d always assumed that ‘user’ and ‘person’ were synonymous, so a person who uses a product is always the same user. But maybe the goal of a product is to take a person from being one user to another. Same person, one user one day and a different user the next day. This is a far more interesting way to think about people and users. People are changeable. Users are just states they are changing from and to.
Idea development
There are lots of frameworks and processes for new product development, but I’ve yet to find one for ideas.
Something I try to do when exploring an idea and understand whether it’s worth turning into a piece of work is to start by assuming we shouldn’t do the work, and then try to prove myself wrong. I look for data that suggests doing the work would be valuable. When I was explaining this to a colleague, they referred to it as trying to prevent confirmation bias, which I think is a good way to look at it. It stops you wanting to work on your ‘good’ ideas.