A way to add/organize content in a way that is less rigid, more organic and frictionless (like RemNote)

WHAT DO YOU RECOMMEND

A more freeform/intuitive way to build my knowledge base by being able to easily/quickly nest/rearrange “Objects” around as needed. Structures and categories emerge naturally, and can be changed/improved with no friction.

The short video below is an example of how this can be done in RemNote.

Some notes:

  • In RemNote, each bullet (which they call Rem) is the equivalent to an Object in Anytype. They can be nested, but also opened as a “page” on their own (as I show towards the end of the video)
  • In the video I intentionally avoid using any other RemNote features - focusing on the workflow for adding and organising new basic textual content
  • I also avoid anything too complex - just some very simple dummy content

HOW COULD IT BE DONE

Not sure how the behaviour I show in the video could be achieved in Anytype in its current form (or even in its intended form). As long as something similar to what I’m showing can be achieved somehow, I’m happy.

It seems to me that the way Anytype is currently built isn’t well suited for the workflow I show in the video. Currently AT isn’t very granular - Objects/Pages can’t really be nested within each other. There is a block type for nested content (Toggles?), but it’s quite limiting. To achieve what I want, I think Anytype would need to have greater granularity, down to Block level at least.

REAL WORLD USE CASES

The way Anytype expects users to add content is very limited and rigid. You need to decide what is an Object (or Page), and what are Blocks. That’s not how our minds work. I want to be able to just add notes/content, and gradually organize and build up a hierarchal structure/logic around it - everything is an Object (considering the current definition of Object), even down to Block level. And if needed, move everything around, without any friction.

A topic that is minor one day, tucked inside my Knowledge Base, can become a top level topic the next day. And this should be doable with very little effort or impact.

RECOMMENDED ALTERNATIVES

At the moment, the closest I could get to that in Anytype is using inline Sets (when these are available), as these would more or less allow objects to be living inside other objects. But that’s a very limiting version of the behaviour I show in the video.

Sets are an amazing and very useful way to display content. But I wouldn’t want all my content shown this way.

ADDITIONAL CONTEXT

RemNote is one of the most flexible apps around for creating/organising content. Problem is it doesn’t look like they will add database (Sets) view so soon. Also, they don’t have a cheaper local-only version. Anytype has everything I need, except for the agility/flexibility in adding/organising content shown in the video.

Related post:

15 Likes

Thanks for the elaborate write-up! I know you are the author of this topic, too, but I think it is important to link it here to stress the importance of the current difference between Object vs Block:

2 Likes

Thanks!

I’ve been avoiding linking to that because it’s a bit messy. My thoughts about that have developed and matured, especially with all the conversations going on in the forum. I need to do an update on that eventually…

Edit: but you’re right! thay should be linked here too.

1 Like

So if I understand correctly, this appears to be the combination of:

Block parent-child basically exists, although it may not work exactly as needed to enable Rem-like functionality fully. In particular I’m not sure if the “parentage” has data model consequences beyond page layout. But it should be something that could be represented in an “expanded”/focused view, I would think.

“Focus”/expand block deserves its own feature request, if it doesn’t already exist. I’m don’t recall seeing it, but forum content master @sambouwer might know better than I. :grin:

And as noted, toggle/collapse as a function of every block already has been requested, accepted, and is supposed to be in dev (:tada:). :slight_smile:

Did I miss anything that would be required to make this work within Anytype? Any function that those 3 capabilities would not enable?

3 Likes

The features you mention are great, but for the workflow I mention in my post, I think these wouldn’t be enough.

In Anytype, when you want to enter some information in your knowledge base, you need to first decide: should this be an Object, or a Block within an existing Object? That is much more friction than the workflow I show in my video.

What makes this work in RemNote is the fact that there is no imposed hierarchy (like the Object > Blocks in Anytype). In RemNote, everything is a Rem. Even root Rems aren’t “special”, since they can always be moved into the hierarchy as needed. There is zero friction in entering, editing and organizing content.

Elsewhere someone mentioned the possibility of adding a feature to Anytype allowing the user to convert Objects to Blocks and vice versa. To me that would be very convoluted, as it would be a workaround to make Anytype work in a way that (it seems) it isn’t inherently built to do…


By the way, I hope it’s not an issue that I refer to RemNote (as well as other software) so often. I think Anytype is amazing. Which is why if some other similar Knowledge-base software has something that I really like, I make a FR here. Hope I don’t come across as ungrateful or difficult :stuck_out_tongue:

8 Likes

tldr: Adding to my original post, I’ve been noticing a trend towards this block-based data model across other similar apps. Since this is an essential feature to my workflow, I am starting to reconsider if Anytype.will be the app for my particular case. Especially as other apps have been catching up in terms of database view capabilities.

Even just a year ago, the only apps I knew that offered a block-based data model were Notion and RemNote. Due to Notion not having offline mode, it wasn’t an option for me. So I adopted Remnote.

Remnote still missed the ability to query and display my content as a table (or any other view) - aka database view. This is why I started gravitating to Anytype, which is practically built on this concept.

Right now, there are a few apps that seem to have, or to be close to achieving, both of these features that I need - block-level granularity and database capabilities.

Tana seems to do it all, though it seems to be online only, and pricing is prohibitive for me.

RemNote claims to be working on database views for it’s next version, though hard to tell how far off this will be. It’s what I use now, since to me it’s the best option (free plan, local only). Database view will only be available in the Pro plan, but the pricing is not that bad.

Other notable mentions are Affine and
siyuan, though hard to tell how long until a stable version of them is released.

Of course, none of these apps are perfect. If you ask me, I’d much rather have Anytype adopt this block-based approach, so I can continue investing my time on it as we wait for Beta. However it doesn’t seem to be likely, at least not on short or even medium term.

Of course I respect Anytypes decision to focus on a different data model approach. In a way I guess I’m posting this just to explain that this wave of new alternatives is why I might be a bit less present in the discussions in this forum (despite people here being so awesome!).

Would be happy to hear any of your thoughts or opinions on this!

9 Likes

I think Tana in particular demonstrates a good approach to this kind of flexible + structured approach, that potentially solves many problems. It feels both more flexible than “everything is an object” and more intuitive than Relations, Types, etc. are right now in Anytype. So regardless of whether it is a tool that any one person can use, I think it’s a good one to watch some videos on to understand the approach they are taking, and help inform the development of other tools (including Anytype).

The nature of this PKM tools market is such that new ideas tend to be copied quickly. I suspect that Tana’s approach, while innovative, will be available in a local-first, and perhaps even open source app (whether Anytype or something else) within 1-2 years.

8 Likes

I agree about the cross-pollination of ideas. But something so structural (such as a decision on which data model to adopt) is much less likely to be changed. Unless a full refactor is done - but Anytype has just done one of those.

1 Like

I agree that we need a way to be more flexible with blocks, which includes being to tag a block and that we should be able to live search a tag. The way https://tana.inc is able to create versatility with blocks - which they call nodes and tagging, is phenomenal. It allows for setting up live searches, which in Anytype is the equivalent of sets. but it can then be done at a much more granular level. In a way, it is not true that everything in Anytype is an object. A tag does not behave as an object. I am aware that the tag function will be redesigned and I am looking forward to see the result. Ideally I would like to be able to tag a block and create a set in which I can gather all the block with that specific block. I wonder if Anytype can also replicate the idea of supertags where any tag can have fields.

I recommend everyone to check out https://vimeo.com/752422100 and Tana Fundamentals 06 – Supertags and Search - YouTube because it shows what is possible in terms of live searches and presentation formats. I think using this database format that Tana is using will be the killer feature for many PKM apps. If Anytype can implement this, it will be very powerful and really be a comprehensive OS. I believe that this will be the killer feature for Anytype. I know that Remote is also coming online with a database that can do similar things as Tana in their upcoming version 2.0. The problem with Tana is that it is all in the cloud and they have no offline version.Anytype is far more secure and preferable. Remote also has an offline version.

7 Likes

Exactly. Though, terminology aside, this seems to be by design. I don’t think AT is interested in this “everything is a X” approach. To pull this off you need to build your entire system on top of this more granular data model. It doesn’t look like this is possible in the way AT is structured (Pages, or Objects, as the building block). And since they recently did a huge refactor, I think it’s unlikely they’ll do big changes any time soon.

By the way, does RemNote have a ballpark date for 2.0 already?

1 Like

I am not aware of a date for the Remote 2.0 version. The last time they did a major upgrade quite a few things went wrong. I think they will be more careful now, which means perhaps it is half a year away.

I wonder if it is not so easy as to give every block a unique ID which would make it in essence an object. Then you can manipulate every block through live searches. If multiple blocks would have a certain word in it, I could search for this word and all these blocks would show up and this could be a set.

What Tana teaches us is that a Tag can also be an object.

2 Likes

Maybe. I’m no a developer, so I’m just speculating.

For example, if they are to make each Block (and every other thing) the equivalent to what is currently called an Object, that would mean an exponential increase in the number of Objects. Suddenly their system will need to be able to tackle 10-100x more objects…

Anyway, I hope you’re right and I’m wrong :smiley:

What I need: Use case:
Ability to create a live search for words in a text that are not defined as an object.

My use case:

I gather quotes, from different authors. I want to have a set that gathers all quotes from 1 particular author and can present this in list view or table view. The quotes I gather from my Readwise highlights. In Anytype these quotes may be part of different objects, e.g. daily reflection, Thought, etc. At the end of the Readwise highlight the author’s name is in brackets followed by the book title e.g. (Steve Smith, Just a book).
I cannot easily do this at all in Anytype. But in Tana is can do the following:

  1. Put cursor anywhere on a page
  2. press CMD-O which brings up a search bar
  3. I select Find nodes with text “”
  4. I type Steve Smith
  5. Press return,I select “as list”
  6. I click return and it presents me with all the quotes as a live query in list format.And the Author’s name was not tagged in anyway, nor was it defined as a separate object as Author. It was just two words.
  7. If I now add another quote by author Steve Smith, it will automatically add this to the Live query
  8. I can rename the live query however I want to. I can manipulate the views like in Anytype.
  9. Set this live query once, and forget it, but it will always update.

The above sequence is a lot less friction than in Anytype where I would have top go through a lot of hoops to maske this happen. I need to do a lot more thinking. While these are 9 steps, it takes about 1 minute to accomplish this.

Since users will gravitate to less and less friction and more flexibility because knowledge bases evolve over time, I think we need to critically look at the architecture of Anytype and work towards the flexibility to be able to find individual words that are not necessarily defined as a tag or object and be able to save them as a live query.

At the moment you can do this already in Remnote, but you can not manipulate it from list into card, or table. In essence it becomes a search portal. With version 2 I think Remnote may be on par with Tana.

5 Likes

Here’s another app that seems to have the same type of flexible data structure: Legend app

Based on recent information from the team, it doesn’t seem like the approach I suggest in this post will be pursued by Anytype, at least not anytime soon.

Could someone from the team maybe comment on this, if there is any chance something like this might be considered in the future or not?

2 Likes

Hi @qualquertipo I personally really like the overall idea. There are even really neat prototypes like this designed. For now you can transform any block to object via turn into object action and we want to make it easier as well as some kind of easier preview of object content.
For this year our priorities is quality/reliability + polishing app for users use cases + collaboration.
Anytype is not an outliner and more a block based app and we don’t have any major plans on changing how Anytype works this year. If we will understand it is crucial for major use cases we support to provide this level of flexibility we will definitely do it.

4 Likes

Link is private :point_up_2:


Re: via @philoup

1 Like

This is a nice idea, but it’s not really what I am talking about. I’d like much less friction than that - blocks and objects as one in the same, as I describe in my original post. I’d like to be able to do what I show in the video I made (create and move around blocks/objects freely).

I am assuming you meant “not” here (instead of “now”).

Anyway, it doesn’t seem like its realistic to expect Anytype to become something like I describe in my original post, as this would require major refactoring… I understand and respect this is a design choice. Just wanted to confirm that!

Thanks for replying.

2 Likes

(Posting in 2 topics because I am not sure which one of those is actually active…)

I also think it would be very handy to be able to set relationships for an object directly inline, without needing to open the page for that object.

Fibery implement this in a wonderful way:
Uploading: 2023-03-08_22h58_14.mp4…

2 Likes

:wave:t3: Hi @qualquertipo

Maybe you already know this, but I just wanted to let you know. I tried RemNote today and there is a free version that can only be used locally, as well as the ability to add databases. :nerd_face:

To me, AnyType looks better aesthetically, so I’m waiting for the next updates which will enable self-hosting. :person_in_lotus_position:t3:

1 Like