Derek Sivers: Write Plain Text Files

Take a look at the export options–there are a lot of them.

And for those so inclined, they claim there are open source libraries that theoretically allow you to do the conversion yourself if you lose access to the app. Obviously if one is planning on that as a future-proofing strategy it would be best to download / test that before locking all of one’s data in. :slight_smile:


A major problem with long term record retention is the overhead of validating that your important old documents can be opened by newer versions of the software that produced them. That overhead grows with time, as long term records accumulate. If the information stored in documents for long periods is important, then that kind of validation is equally important.

The likelihood of maintaining compatibility decreases while the difficulty of extracting information increases relative to the amount of metadata that accompanies the information in question. There’s also a tendency for the importance of the metadata to decrease over time: Nobody (or at least very, very few people) care much about the original formatting of Shakespeare’s work relative to the content.

The more important the longevity information, the more important simple representation is. If I think something’s going to be needed a decade out, I almost always opt for plain text (if possible), because I know that I won’t keep up with any kind of access validation.

On the other hand, my CV lives in a Word document because formatting is important and it gets revised at least once a year.


Just checking, the oldest Word document I have in active use is 23 years old, and opens fine in the current version of Word.


Add to that the fact that the potential for unrecoverable data corruption would seem to increase with format complexity. Corrupting one of the early parts of a zip file can potentially render the whole thing unextractable, whereas the exact same disk write error in a folder of text files would likely only take out a handful of documents.

1 Like

@svsmailus, I may disagree with what you say, but I will defend to the death with a blog post your right to say it. :slightly_smiling_face:

(See, Markdown can do Strikethrough!)

Well, this spurred more discussion than I expected so far.

I’m not really concerned about Word docs, personally (I only ever use them from clients).

I am much more concerned about all these note-taking apps that lock things into their own systems. Are my notes in Ulysses really going to be all that accessible in iCloud after Ulysses shuts down? (I know I could use external folders, but the train left the station for that option a long time ago with my use.)

What about Notion? What happens when they shut down to the notes I keep there?

I tried Craft as well for a while. It doesn’t meet all my needs for a notes app, but I have tried exporting my notes out of it. It’s still painful. One note at a time, clearing up weird formatting issues, etc. It’s not “plain text.” It’s Markdown, done the Craft way, with their own formatting baked in (pre-baked headers and the like that you can’t change). More power to those who like Craft, it’s just not comparable to plain text.

What happens to all our stuff in all these web services once the web services go away? At least .doc files are often local to my machine, or at least visible in a directory structure somewhere.


Would you not consider “vanilla” Markdown to be “plain text”? Or is it just the Craft extensions that put it in that category?

Yeah, that’s the rub. I would assume that the files stored in iCloud would still be there, but I wouldn’t want to bet on being able to get the data out. I think the standard thought is “if they’re going to shut down, then you can export your data”. But honestly, that seems both (a) overly optimistic, and (b) a bit challenging.

When I used Ulysses I used Markdown to avoid that problem, and at least there were copies of the file in my iCloud. But for services where the data is mostly or completely online, yeah - you’re one unpaid server bill (or more realistically, one server issue that they can’t recover a backup for) away from your data being gone.

If I were in that situation I’d be considering whether there was some easy way to auto-export the data. But I’d be cautious about getting into that scenario in the first place. :slight_smile:

1 Like

This was one of my biggest concerns using web based services. When I ran AWS servers I felt the bigger risk was the account being suspended or closed (due to an accidentally missed bill or a system wrongly interpreting use of the system as hacking/spam) than technical data loss. Even suspension for a few days would have been very difficult. It was expensive to have non-AWS backups rather than relying entirely on their (technically robust) backup strategies across availability zones, but kept me sane!

Similarly, without the ability to have locally usable files automatically backed up, I would be reluctant to have critical data on a cloud based system, even if you can potentially export. You typically won’t get any notice of an account suspension or deletion.

1 Like

It’s probably more likely that what you write today will mean not much to you in a few years. In the decades I’ve been following and participating in software discussion forums I’ve seen recurring angst about “lock-in” but rarely see much discussion about “sugar, I’ve been burned by lock-in and I can’t get my stuff any more”.

@OogieM’s unfortunate case, of course, disproves my point

Yes. I read the article. Sounds like good advice, for the writer in particular and for others who have similar needs and preferences.

I’ve made my living as a writer for something approaching 40 years. I write journals to organize my thoughts and for the sheer pleasure of expressing myself. I’ve written long and short fiction and many, many scripts for the stage as a hobby. And since young adulthood I’ve been writing indulgently long letters and e-mails to any friends and family willing to endure them.

Serious consideration of format and software has been rare, usually reserved for those times when I know somebody won’t pay me if I don’t comply with their particular requirements.

And I’m more than certain that format and software rarely, if ever, play a role in distraction or focus. Solving distraction issues always begins — and usually ends — with an honest assessment of my relationship with or dedication to any particular writing task and how well I’ve prepared myself to make it happen.

I’ve lost more of my writing than I could ever find time to worry about.

Some of it in plain text format.


I appreciate your comments and perspective, quite helpful. I am a bit confused, however. Based on your extensive experience, are you saying you are or are not an advocate for plain text? I’m not at all challenging, just not quite sure where you landed. “Its me, not you!” :slightly_smiling_face:

Also, I agree. I don’t think the application is usually the problem with Focus. For example, both Word and Pages have good focus options to remove the distractions, one doesn’t need a plain text editor to write distraction free.

I was merely making a simple statement based on the fact that some on this forum do advocate or advise (take your pick of words) for the use of plain text whenever possible whereas others don’t believe it fundamentally matters, that’s all. :slightly_smiling_face:

I had to think this one through for a good couple days, it really stuck in my craw. I took a look at my use of file formats and applications, I’ve got a pretty good mix. Lots of plain text markdown, lots of OmniOutliner and OmniGraffle files. Hundreds of entries in Apple’s Notes, years of journaling in Day One.

I considered what it would take to convert everything to plain text, and what I would give up. The ubiquity and simplicity of Notes and the historical metadata of Day One being the biggest issues I’d have. After staring at a pile of plain text files and imagining myself relishing the purity of my Mac being nothing but text I came back to the reasons I use a Mac and not a PC running Linux.

It’s very easy for technologists to fall into a trap of fetishizing this imagined technical purity, and to overlook the fact that the computer is a tool. It’s not about “how” you do a thing, it’s about actually doing the thing. My job is to know things, when it comes down to it, so whatever tool helps me to know things the best benefits me. Are there risks involved in data silos? Absolutely, and I take care to work with applications from companies I trust, and occasionally check on what it would take to export all my data and move to something else. But for me, I value high quality software that I trust and that is a joy to use.

So, hey, if you want to go all plain text… go for it! Have fun! But if you don’t, and you actually enjoy using lots of different software, don’t feel bad about it! The Mac is a great machine, and it can do lots of things. Explore and enjoy.


I think most of my mediocre dramatic writing would leave you similarly scratching your head. I’m saying it doesn’t matter — to me.

The writer’s perfectly reasonable choices don’t apply to me, despite the fact that writing has fed, housed, and clothed me for a long time.

I’ve done a pretty good job in recent years of not responding every time I read someone on the internet telling us all what we HAVE to do. But every once in a while I enjoy writing a rebuttal. (And then, sometimes, explaining it because I wasn’t clear the first time.)

1 Like


And each new upgrade/re-verification gives me a chance to re-evaluate the need for those long term documents but in most cases the few nuggets I really need that cannot be found elsewhere are valuable enough for me to usually plan on several weeks worth of major amounts of time to convert the archive and rebuild the backup string.

In the really long term, arecheological data, Shakespeare’s work that may be true but in my personal archives the metadata generated at or updated at each upgrade increases the value of the collection as a whole.

That’s because when it happens the memories are so painful that those of us who have experienced it are often loathe to rub salt in the wound.

On a more serious note mymost recent data lss of files stored on suppossedly archival 100 yr old CD-ROMs that are now totally unreadable only spurs me on to make sure I am religious about my archive verification at least once every year or 2.


I’m talking mostly about informaiton stored as freeform text (writing) and metadata in this context is all the extra stuff required for formatting.

The value of metadata when it’s used to enhance the semantic content of data, or describe relationships between data doesn’t degrade over time relative to the data itself, or doesn’t do so in the same way as when it’s only there as a container or as asthetic formatting information.

ah that makes sense. I was thinking of more traditional metadata, data about your data not the presentation of the data.

This has been an interesting discussion! Of course it does highlight the adage that has already been mentioned, of using the right tool for the right job.

My difficulty with plain text is that usually for me it needs to become text with headers, lists, illustrations, footnotes, bibliographies and the like. Plain text just doesn’t do this or doesn’t do it well. Someone will give a shout out for markdown, but vanilla markdown is limited and the proliferation of markdown variants is tiresome. I had a major headache getting my data out of Ulysses as I’d used their XL Markdown. Of course markdown itself requires an editor that understands it’s syntax. I don’t see this being massively different from any other markup language. Markdown’s strength is it’s simplicty, but that’s its greatest weak point. Simplicity rapidly looses it’s appeal when you want a diagram with a caption where the text flows to the right of the diagram or a complex table. In the end everything becomes a workaround and more importantly more steps are added to your workflow. I’ve yet to find a markdown to PDF converter that handles page breaks well and headers and footers.

In the end I would agree that storing your data as plain text seems good, but once work commences on the text to add meaning, emphasis, clarity and explanation doing it in plain text never works for me because of its inherent limitations. Also once you have the finished article or paper, returning it to plain text will lose a lot of meaning. I also prefer my developed thoughts not my initial plain text thoughts.

In the end this returns us to our individual contexts and making decisions as to what we use based on our needs. The benefit of today’s software is that many allow export. MS Word allows, text, RTF, HTML and PDF, surely that mitigates any form of lock-in? My main reason for using Word is not that I’m a lover of MS, but that Word’s ubiquity makes it the obvious choice in many cases.


Begin the journey with an idea of the destination and the vehicle needed to arrive there. Don’t rent a Bird scooter with an idea to get to Antartica.