633: Workflows with Nick Milo

And attachments can’t be tagged (or have any other metadata), which seems like a step backwards in connecting thoughts to me.

In something like devonthink, I could have a tag called Aristotle, perhaps nested under Philosophers, and apply that tag to any of my notes, audio books, pdfs, etc.

I would say on the software side, the notable features for revealing connections are:

  1. Very low friction linking – previous generations of this sort of tool did allow you to get links to individual items, but you usually had to go to the item you wanted to link to, click on a menu, copy a link, format it, and paste it into the document where you want the link. In Obsidian (and other newer tools like Noteplan, etc.) you just hit square bracket twice, start typing, and it will autocomplete.
  2. Backlinks – Every document has a list of the things that link to it, allowing you to choose whether that connection should be bidrectional.
  3. Unlinked mentions.

Of these, I think #1 is by far the most important, followed by 2 and 3.

That said, the real power here doesn’t just come from the software features, it comes from the effort the user puts in to initially set up these links (which is why the low-friction linking is so important) and then going into your database and exploring these connections through the links and backlinks (and maybe unlinked mentions).

2 Likes

Be careful. It’s only free fir non-commercial work. If you’re using it for commercial work, then it’s $50 a year.

I’m also not sold on Obsidian.

I think it’s because I’m a one big page per topic person. This stems from me usually writing out notes by hand as I read. I could have an additional processing step to transfer those as small linked notes, but I just don’t want to manually manage a bunch of metadata to do it. Should I change this system? Probably, but it’s worked for me.

I also just really don’t like markdown as syntax. I prefer org syntax instead. I tried logseq since they support org, but couldn’t handle that everything had to be a block.

No matter what you use I think there’s two key components to any modern notes system:

  1. It must have very good search. Life is too short to be managing directories and file names in 2022.
  2. All systems are worthless unless you review the notes. I think people get so hung up on the Zettekkasten linking system that they don’t realize the power was in the constant review.

To get a bit more metaphorical than my previous answer: a PKM system is like a garden. Obsidian gives the user the soil and gardening tools, but it’s up to the user to plant the seeds and nurture them.

4 Likes

I find the comments about Obsidian not quite being the be all and end all….expected.

We’ve been through this cycle a few times where a new app drops and catches the attention of all the podcasters. It’s faster, it’s more robust, it allows you do to X more easily and Y is on the way. There’s a honeymoon period and then as the user base grows you start seeing the “I love this app but I wish it had ____” posts, then the “it’s great but it’s so slow”.

It’s hard to not be tempted by the latest an greatest. Some may disagree, but the most recent cycle of this would be Notion. It’s amazing, I use it for everything, it’s a game changer —— it has no offline mode, it’s slow, you can’t really get your data out of it easily….and on to the next one.

There’s so much bouncing around in the space. I’ve gone into Obsidian fairly deeply since leaving Evernote and I’d be lying if I said I didn’t occasionally think - “man, life was so much easier when I could just dump it all in Evernote with tags and move on with my day”.

1 Like

It’s also worth noting that DEVONthink’s default method of storing data uses a database rather than a folder structure. So if you just throw stuff in DEVONthink, even though you’re creating groups and making things that look like folders, what you really have on the back end is a relatively flat folder hierarchy. I wouldn’t want to put one of those systems back together if it crashed.

DEVONthink of course can also index an existing file folder, but that has its own set of tradeoffs.

Obsidian’s data store looks exactly like your folder in Finder.

It’s a rather technical point, but it matters to some people. :slight_smile:

No, I’d say that’s a pretty big difference. It is one of the main reasons I quit DEVONthink in favor of EagleFiler.

1 Like

I used Evernote for years and relied on search. When I left EN I dumped everything, except my sensitive data, on Google Drive and still rely on search. Anything not on GD is in EagleFiler. And things are still pretty easy.

I do miss the Evernote feature that included my data on the page when I did a Google search. But saving the extra minute or two it takes to locate my files on Google Drive isn’t worth $100+ a year.

1 Like

I use DevonThink to archive and index any PDFs that I read. Basically if I read and want to remember it, then it goes in DT. When I want to take notes they go in Obsidian.

Hope this helps
- Another - Mark

2 Likes

The discussion of dopamine addiction reminds me of this article in Forbes about “over motivation” Too much motivation can produce bad decisions. The author demonstrates this aspect of brain science with the movie American Hustle.

1 Like

I love that Stephen got to teach the sherlocking concept to Nick.

If you want fuzzier note relations, Ryan Murphy’s plugin (edit: now with working link!) grabs and links similar notes from DT when you have DevonThink indexing your Obsidian folder.

There’s also GitHub - tokuhirom/obsidian-2hop-links-plugin which tries to show you what your links are linking to. It unearths some interesting indirect connections, but visually it’s rough and interacting with the display is disorienting. I think this could be worth rewriting with a few different features.

Autogenerated links based on content (much closer to DT style) is the holy grail–if there’s a plugin doing it well, let alone at all, I’m unaware.

As for hype cycles, at this point Obsidian the nice-looking plain text wiki and Obsidian the highly extended database backed by a large plugin community don’t have a lot to do with each other. I do agree some don’t end up liking the tradeoffs necessary to store most of Obsidian’s data in plain text.

1 Like

Are you referring to plain Obsidian vs. Obsidian-with-Dataview plugin here?

2 Likes

Pretty much, yes. :slight_smile: Well put.

1 Like

I keep hearing Obsidian fans saying that one of the biggest benefits of using the app is that since everything is markdown, it’s future-proof and can be easily moved out of the app if necessary. (But of course, other apps have long included markdown and plain text files in their systems, and can be easily moved, such as DevonThink.)

But isn’t the real “killer” primary benefit of Obsidian supposedly the linking between notes?? And that’s based on Obsidian’s own unique system, right?

So if you move your markdown notes out of Obsidian, you lose the primary value of having them in Obsidian in the first place, which is the linking. So how is that future-proof?

I guess you could hope that someone would develop some way to preserve the linking, but that could be said of any number of linking apps …

The basic double-bracket linking syntax is common across certain other apps (Noteplan for instance). Some of the more advanced syntax is not, however.

I index my Obsidian vault to DEVONThink, where the links work fine on Mac but not iOS. This is enough future-proofing for me.

It’s helpful to consider the context in which phrases like “future proofing” are used.

No software can offer functionality in a data export. In other words, if Obsidian magically creates links every time it sees [[…]] that functionality won’t be inherent in your data.

But with data like this, “future proofing” generally does mean that there isn’t any proprietary data relative to the program that’s required to re-assemble your data in the form you would expect.

Consider an example.

Let’s say you put your notes in DEVONthink. You make files and folders for clarity. Interlink them. And then later on, you go to move your data out.

If you used DEVONthink’s item linking (i.e. you linked using the item ID in DT’s database), your links will go away because that index is stored nowhere in your data - it’s proprietary to DEVONthink. And furthermore, even if you imported all the same files to a NEW instance of DEVONthink, the IDs would be generated anew and the links wouldn’t work.

Of course if you used the double-bracket syntax for linking, you’ll still be good so far.

But what you’ll discover next is that the file/folder hierarchy you’re seeing in DEVONthink doesn’t at all resemble the folder structure on the disk. Your double-bracket notation could be picked up by any other app, but everything else will be an absolute jumble due to the fact that the DT interface doesn’t resemble anything that’s actually happening on your disk.

With something like Obsidian, you use the double-bracket notation to link. You make files and folders to organize notes. And if you delete Obsidian, your folder on disk looks exactly like it did in the Obsidian interface.

No software can offer complete export of functionality. But if you’re working with files in Obsidian, an export of your data looks exactly like you’d expect it to look. Files, folders, double-bracket notation, etc. There’s nothing about your data that’s stored away in a proprietary format that you can’t access. It’s all there in the files.

Does that make sense?

And of course there are reasons DT does things the way it does, and potential benefits of that method over Obsidian’s. Many times, giving up “future proofing” yields significant benefits for present-day usage. But Obsidian’s “future proofing” is definitely better.

2 Likes

I have no doubt that my pdf files will be 1. readable long after I’m gone, and 2. of no interest to anyone.

But when the question of what file formats are preferred for long term record keeping, I check with those in the business.

1 Like

The thing, of course, about Markdown is that it’s so dead-simple as a format that it’s effectively equivalent to ASCII. It’s basically human-readable as-is, and writing a Markdown parser for any system that uses ASCII text is a trivial task.