If you’re great at queries and can constrain yourself to having certain canonical blocks for a concept, you technically could use block references instead of page references. The trade off is that filters would be harder to use since you can only filter by page refs @RoamResearch
If I were to do that, then all of my most important notes would just come with certain text in the block to make it easy to autocomplete. For example, all of my open questions have “open question” in the text of the block so that ((autocomplete is searchable.
What’s nice about this approach is that it gives you autocomplete but for whole thoughts, whereas it’s cumbersome to put more than a sentence into a page name. Actually, I might do this with my “evergreen note titles” too... use them as block refs for more robust autocomplete
If you want to use blocks at that atomic of a level while also using them as an organizational layer for your notes, then the words within the block are key for resurfacing via autocomplete. In the example above, I instantly filter down everything by writing in ((open question))
The magic comes then from combining keyword/key phrase level page references with block references here. The main blocks of interest are embedded within block hierarchies with page refs that make them resurfaceable... allows you to find things even when words aren’t in the blocks
Just as a reminder- you can query block references just like you can query page refs. For queries, all that matters is easy autocomplete to pull up what you’re looking for. QT shows an example. I’ll query a block ref to a specific open question all of the time.
The biggest obstacle to using block references as more verbose page refs is that, if you just think out loud in Roam, you’re way more likely to end up with many blocks that say the same thing vs. many pages that refer to the same idea. So which of the “duplicates” do you query?
That’s a big reason why I still recommend using page references for keywords or terse statements. It’s just easier to avoid duplication that way. It’s most likely true that you’ll have way more blocks than pages, so page refs let you have a smaller selection to autocomplete from
There’s a lot of literature on autocomplete. @worrydream says that brevity is irrelevant in an environment with good autocomplete. That’s why placing keywords in the core blocks you want to reference is important- lets you use verbosity to your advantage to filter down easily.
Roam could enable this further (cc @thepericulum) with advanced search terms with (()) inline searches. For example ((open question -resolved)). Or making that possible in the search bar, with a keyboard shortcut to paste selected page or block ref to your cursor
I found this block by searching ((quote Javier)).
Reminder: zoom in on a block (ctrl+o) to filter block refs, but you can only filter by page refs, not block refs. Queries remove those limits. This is why my advice in this thread doesn’t apply if you rely on filters over queries
• • •
Missing some Tweet in this thread? You can try to
force a refresh
One challenge with explainers is that, no matter how clear they are, many people just skip them.
Something I love about feedback loops is that, as long as people are using the product, they'll learn how to use it. Skipping feedback loops means nonusage.
Explainers feel appropriate for more complicated concepts... but does that really teach the nuance better than trial and error that clearly shows the outcomes of your behavior?
An explainer might help people feel more comfortable getting started and triggering the loops.
It would totally be possible to use @figmadesign (albeit hacky) as a personal knowledge management system given certain user behaviors & maybe a few new plugins 🤔 It wouldn't be nearly as powerful as say @RoamResearch or @kosmik_app but it would outperform traditional note apps
It has a lot of the same limitations of text vs. GUI I describe in this thread. Ultimately, if an idea doesn’t need to be expressed with visuals, it’s just really hard to beat the speed of typing. That being said, doing it visually leads to me thinking through it differently.
An alternative view on what a visually created PKM system could look like in this thread, as well as musings on the limitations of Figma that could perhaps be addressed via plugins
I love this part of the @worldbrain vision. You can create a collection for web pages. Others can follow/join if they are interested in the subject. You can then set the rules for participating in an online conversation between community members that spans across many web pages.
Imagine this... I want to create a collection for all of the behavioral science articles I read, and know a few other people who I trust to add good articles to it. They could be set as co-curators, or just follow the collection. We’d then talk about each article on the article!
It’s sort of like if we set up a forum for ourselves to discuss articles, except the forum follows us around. It’s like placing my own comments section on someone else’s site.
Imo ABA therapy isn’t effective because of the power of behaviorism, but rather because you have a therapist paying very close attention to what makes an individual kid tick. Of course that will work, but you don’t have to do verbal gymnastics to avoid attributing internal states
This is an unnuanced take about an intellectual gripe I have that doesn’t really matter
Right, it’s individualized care and close attention/analysis. Behaviorism as a philosophy isn’t really necessary here 🤷♂️
Figma totally just leapfrogged Whimsical for me. I see no reason to use Whimsical ever again.
This is true. I was mainly using it for quick flowcharts. But once you set up a few components for yourself, spend an hour learning the basics, and install Autoflow, the flowcharts become about as quick to make as in Whimsical
So the way I see it, I can spend that hour learning and do a little setup and get access to a tool that does everything and more, allowing me to increase my skill level over time in something that rewards that growth. Whimsical has a lower floor but also a way lower ceiling
This right here is a huge area of exploration for me since working on GuidedTrack, realizing that @worrydream had already written about live programming environments in the context of learnable programming, and even contemplating things I could build myself
Big open questions: what broken glass to get started could be offloaded to design choices? How can languages and IDEs be designed to be approachable and understandable, providing strong feedback on the impact of your actions, and encouraging you to engage feedback loops?
Part of what’s appealing about learnable programming is just how hard it is. I’ve worked on 6 onboardings so far, and working on a programming language/IDE (GuidedTrack) was by far the most challenging.