Will we have a DT 4 in the near future?

Not all of us see this as a feature. I want to use my information, not teach my computer how to manipulate it.

My (admittedly brief) experience with Obsidian is that it doesn’t scale to anywhere near the quantity of information that I have in DT.


Haha! Well, our “bad business practices” have kept us solvent and sailing steady for 22 years now. :thinking::slight_smile:

Here’s to 22 more! :beers: :beer:


Very true. As I said, I’m using DTP mostly for storing emails. Like, millions of them.

At the same time, then again, I am again watching my iPhone as it takes days to download a year’s worth of emails to DTTG. That on a network where both the webdav server and the iPhone are on, and I get about 450 Mbps throughput to my iPhone easily via Wifi. Once they’re downloaded, it’s cool. For some reason, my existing databases on the iPhone didn’t synchronize anymore after I had moved one year’s worth of data to another database (like, move 2022 out and start with 2023 only for my current mail database). It was just busy showing a kind of a HUD message for every single filename it probably didn’t find anymore (of course with just some random numbers, so not only annoying, but also unusable) - that I ended up just deleting those databases altogether on the phone and synching them again.

I’m digressing. DTP is a thing that I keep using for lack of a better option for the specific use case I’m using it for - storage and fast search across millions of mails. I’d never use it for anything else these days.

Oh interesting. And thanks for taking the time to read my article. It’s probably what you say, a potentially very different user base. I don’t give a lot for “millions of users” (I’d hope that DTP had so many, would be great for them). For the users vs. developers, interestingly, the first other Obsidian user I accidentally met in person was the then curator of the Royal Art collection. He was doing all his research on Obsidian and was probably the most technology distant person that was using a computer I had ever seen in my life. And that was back in 2021, after Obsidian was basically only out for a year or so.

So if someone dusting off old posh furniture in Buckingham Palace and Windsor Castle gets to use Obsidian, it can’t be too developer centric. Of course I do push it, but that’s the whole point: I’m actually able to change just about everything of it, whilst in DTP I literally can’t even have my trash folders not using a ridiculous strikeout font (I keep coming back to this, as I got pushback even for asking to have that configurable). I can do some cool automations with DTP, like writing some simple apple scripts to import mails etc., but then again that’s really limited.

I don’t care about price really. And Obsidian’s sharing option I also don’t care about; theres’ e.g. git plugins available, and before they were, I just wrote one myself. So you don’t actually need to “buy” anything from them; as you can imagine, their sharing option via their cloud is clearly a no-go for enterprise users.

But then again, you have the option of doing things otherwise.

A limitation in automation is largely dependent on the creativity and/or level of expertise of the automator. DEVONthink has some of the most flexible and robust automation possibilities on the platform.


Does this guy know the keyboard shortcut to launch the developer tools in Obsidian? If yes, he is probably more tech savvy than the average guy with a smartphone. If no, most of your arguments in the blog post for Obsidian superiority do not apply to him.

Thanks a lot for your approach to LLM. In my book, I’ve elaborated in great length on the unknown side effects on using ChatGPT or any of the popular NLU/NLP-engines for personal and professional work.


I have found very little I can do in Obsidian I can’t do in Devonthink with greater reliability.

Obsidian’s Dataview can be fussy. DT smart groups are not.

I wish DT had a native local node map, but the see-also and links inspectors reveal basically the same information.

Tinderbox does something neither DT or Obsidian will do. Notes in Tinderbox can have sub-notes. In another thread, that’s why I suggested it would be nice for DT to display a group annotation in the preview pane.

All of the above is opinion, of course, and likely flawed.

1 Like

Not sure if it’s been mentioned yet, but elicit.com is a fantastic AI-supported paper discovery and summarization/data extraction online app for scientific literature. Already great and very actively developed.

1 Like

As for Obsidian, I’ve tested both as a standalone App (without DT), and as a companion App (indexing the vault inside DT).

I already had tested Obsidian about 2 years ago, but it never clicked.

So after many months I went back full to DT.
I realised all the features I needed to keep my PKM were already in DT.
Obsidian, on the other side, forced me to some workaround in order to replace the way I used DT.

Ok Obsidian is open…but how open is it?
It has plugins, and they are very powerful, but there is no warranty they will be there from here to some time: when I went back to Obsidian I discovered many plugins I used 2 years ago hadn’t been updated since then.

Furthermore, plugins in Obsidian may be something that will lock you in.
I mean: I use Obsidian because the notes are text files, is it?
Then I start to use plugins such as Data View and similar, very powerful, but…have you ever checked those files with another text editor?

So what would happen if you wanted to move your PKM from Obsidian to another text editor, after using so many and pervasive plugins?

We often look at these things from a technical point of view, but we forget the sink costs.
If you make a heavy use of the plugins in Obsidian, then you’ll not be able to move somewhere else anymore, and it’s where Obsidian risks to lock you in.

That said, I really liked how I was able to manage the notes in Obsidian: very friendly interface and great themes (for the pleasure and relax of my eyes).
I wish DT had something like this!

And thanks to that days I realised how so many things can be converted to a simple text file.
Do you store an episode of your preferred podcast? You transcribe the text and you’re done.
As for that video on Youtube? Maybe you can do the same, or at least you can add a link to the video to the transcription.

On the other side, on DT I know I would be able to move my whole PKM anywhere else with a very low sink cost.

And I really prefer to work with the original files.
It’s ok I can transcribe a podcast, but I like to keep the audio file along with the transcription as annotation, so that I have two versions of the same story.

It’s my own experience, of course, and if people are fine with Obsidian and its plugins, I respect their thought.


The psycho-social expression affinity or disaffection with software is always fascinating. Especially the energy expended in detailing one’s dislike.


That’s planned for a future release.


I agree with the concerns about disruption. Obsidian’s merits are besides the point; disruption doesn’t require just one disruptor.

As enough lower-level pieces become commodified, developers start seeing how to integrate or reimagine the incumbent’s remaining standout features.

In the classic low-cost/low-feature disruptor model, every immediate decision of the expensive/high-end incumbent makes sense and their customers are happy right up until the revolutionized market threatens viability of the business.

@korm was hoping to shut down any further discussion by insulting everyone, but I’m still posting this because I do want DT4 and DT5 to be big financial successes with surprising, trend-setting features.

This necessarily implies that every low-cost/low-feature disruptor model has the potential to bring actual disruption. While technically true, this is not more insightful than the truism that everyone has the potential to become the U.S. president.

One’s …

… come from the likelihood of disruption, not the possibility of disruption. You avoided sharing your opinion about this likelihood, and discouraged others from sharing theirs, by saying that …

How would that help further the discussion?

I was saying that the current trajectories of more open plain text editors combined with ML advances make disruption nearly certain, not just likely.

Disruption is a specific innovation model in which low cost entrants are able to adopt a technical innovation that the incumbents cannot, and that combination ultimately supersedes the incumbents’ offerings even though both types of companies release improvements. Understanding this changes the way incumbents respond to market changes, so it’s not just a truism.

Merely being low cost isn’t disruption.

Focusing on features Obsidian lacks is besides the point because it’s in that category of apps that have the lower cost plus the opportunity. The ultimate winner doesn’t matter since we’re focusing on DT.

Here’s a summary of the concept.

IMO this category of apps includes the vast majority of apps on the market. It is very well perceivable – I would say it’s a truism – that some apps with humble origins could bring evolutionary change to something in the future. Feel free to call that evolutionary change a disruption or anything you prefer. But why should that be a concern?

The “disruptive innovation” paradigm you cited is strictly retroactive. We can never know in advance which app will actually become the disruptor. Therefore this paradigm is mostly meaningless before the disruption has already taken place.

The police know this situation well: Someone in a community is going to commit crimes at some time, but we don’t know who. Your local police is surely not monitoring everyone in the community with even suspicion. With today’s AI capabilities, they might be using a complex and controversial algorithm to determine who are more likely to commit crimes, and direct more resources towards these individuals.

Does Obsidian deserve more suspicion for being a future disruptor than other apps?

  1. YES – certain aspects (a.k.a. features) of Obsidian distinguish it from competitors.
  2. NO – Obsidian lacks certain features critical for disruptive growth.
  3. DON’T KNOW. Then you’re just telling us to be concerned about emerging apps in general.

I hope this analogy makes the point clear: for the sake of discussion, Obsidian’s features, and a lack of such, are of equal importance. I do believe there is a reason Obsidian is mentioned more often than anything else in the forum. There is something appealing, something unique (a.k.a. features) about it, which has been the trigger of this thread in the first place.

1 Like

Some years ago, people were saying that self-driving technology would disrupt the automobile industry. We all know how it went.

Real-life disruption are one-off events that cannot be reproduced. Therefore, a disruption can’t be scientifically certain, or near certain, unless you’re blessed with the gift of hindsight.

Self-driving technology is the opposite; it’s incredibly expensive to develop and it’s not clear yet what technological advantage human-driven cars can’t also adopt for safety and lane-keeping (cameras, radar, ML are all being absorbed at least as fast as Tesla and Waymo can use them, maybe faster.)

Anyway, I get that you’re sour on the idea analysis is possible, so take care. I hope you’re at least rooting for Devon.

Sorry to see you resorting to such phrasing.

While you are aware that self-driving is costly develop, you seem unaware that machine learning is an even costlier technology. In both cases, there are big corporations which happily foots the bill for research and experiment. Once development has borne fruit, mass production would quickly put costs down to affordable levels, as has already happened with ML.

That’s not the key point, however. The divergence between us two is a reflection of the larger debate regarding the legitimacy of generative AI, which, among other things, extracts patterns from apparently unrelated data. The question: Should we actually use these unexplained patterns to predict (or “analysis”, if you prefer) singular future events?

Some people answer with a firm YES. They hold that it’s trivial to use these patterns once they have been generated. If the usage of these patterns would result in a mess, then be it. No one should be held responsible for simply (and legitimately) experimenting with available tools.

Some other people, including me, would say NO. One of our main points is that such patterns are unreliable, because they are either outright so, or have not been proven otherwise. Unreliable patterns may still be useful for provoking further thought, but they should definitely not be used for prediction/analysis. Another point is the lack of responsibility: who will pay the bill if someone elects to follow an unproven path, only to meet a dead end? Should the advocates for this path be held responsible in some way?

Both points of the second group have been reflected in my argument against use of the disruptive innovation paradigm, which is essentially generated through cherry-picking in histories of unrelated disciplines, without regard to the numerous counterexamples available.

That’s definitely the part of it. Right now Devon rightfully looks at what OpenAI et al produces and says it’s unreliable, potentially dangerous, slow and expensive compared to what’s built-in to DT. Their customers appreciate this stance very much. However, Devon also sees what people are valuing about the input/output of an LLM (using LLM as shorthand for all the newly popular tools this decade.)

So, Devon is looking into expanding capabilities to match changing expectations safely and performantly, while all these other companies and communities corralling LLMs/vector searches/expert systems to be performant and provide safe output. It’s a race with some pretty interesting technical and community aspects.

The Christianson model says Devon can look for attributes of the LLM that will obviate features of DT and that Devon is unwilling to give up. This can be for multiple reasons; some examples might be a redesign is too expensive, losing sales or customers short-term is expensive, they would violate a company value.

It also says Devon can look for aspects of DT that will hinder adoption of the low cost entrants’ appealing features, even if Devon is watching closely and plans to enthusiastically adopt them.

Finally, they can look at the financial/operational structure of Devon to see how they would handle changes to the business.

Dealing with those questions earlier means better software and healthier finances in the long term.

Thanks for keeping the discussion going after all!

1 Like