Opinion: development focus should shift towards improving LogSeq's partially unstable core

and_yet_it_moves · May 8, 2023, 1:05am

First, I want to acknowledge that I pay nothing, can expect nothing, and for any functionality in whatever form or shape, I can only be grateful. This is not a complaint, but feedback for the potential benefit of the project.

As a reasonably dedicated user who has observed LogSeq’s evolution and development for some time, I believe that the direction of the development needs to be reevaluated.

The core functionality of LogSeq remains unstable. Numerous bugs concerning the core functionality have been reported a long time ago and are still not fixed. Some rough edges in the core functionality have not been addressed.

It appears that the development is focused on adding shiny new features rather than ensuring that the foundation the tool is built on is stable. While these features might be very useful and provide value, I feel it is a significant mistake to invest effort into them before the core of LogSeq has received a substantial overhaul.

There seems to be insufficient bandwidth to maintain the current features, and developing new ones appears to lead to an even greater backlog. (And it’s worse than a linear effect, due to the combinatorial effects.)

As a single example, a user created a list of <% current page %>-related issues two months ago. A screenshot as of today:

Eight mentioned issues, the first at issue height ~3 000 (today’s is ~10 000). Most or all seem to be actual bugs, at a quick glance. And eight green icons: all eight still open.[1]

I am personally not that affected by any of these bugs (although I’m sure others are) - I provided the above just as a snapshot example.

Symptomatic is also the user’s offer to help out along with a question on how, with no response so far. Again: I’m not saying we can demand such answers from the developers or other involved volunteers. Rather, my point is that I believe not prioritizing answering such questions, and instead developing new features, is a mistake. It has been my view that a general problem with LogSeq is that many reported issues, and not just recent ones, haven’t even received a single comment - usually a poor sign for a project.

A related issue is the weakness of the documentation in combination with questions on this forum asking to clarify undocumented behavior often going unanswered.

I will finally point out an observed improvement - @tienson seems to be more active on this forum now than before, sometimes chiming in and engaging with the users.

I hope, for the best of LogSeq which indeed has potential, that the current development strategies are revised and evaluated.

[1]: They do not seem to be duplicates, or not most of them. It is possible the user selected only opened issues, but there’s not much of a selection effect: searching for closed issues on <% current page %> doesn’t yield many results.

jeroen_ophetweb · May 8, 2023, 2:27pm

Allthough I normally try to not write any ‘me also’ answers, I do want to state that I for one have choosen to use Obsidian for the foreseeable future after almost a year with on and off changing between Logseq and Obsidian. For this exact reason.

In my hart I know I am more of an outliner guy then anything else, but every time I decide to use Logseq I run into some ‘big’ issue that makes it unusable for me. That latest was that renaming a page does not make the required changes everywhere and needs a reindex to make it work correctly again.

In itself most issues are not huge and there are workarounds available mostly, but over time these bulk up and I am in it for the use of the application, not for trying to work around the issues.

As @and_yet_it_moves mentioned for himself, I am also a non-paying customer only using the app on a desktop so I am not entitled to anything, but I do think that in the long run something need to change.

Thanks for the great product you build and I hope you can find a better balance between updates, fixes, documentation, communications and new features. I will keep an eye on Logseq.

Regards,
Jeroen

ThaisRobba · May 8, 2023, 5:08pm

I had to sign up to add my two cents because this has been on my mind for the past few weeks.

I freaking love Logseq. I think it is the solution that most removed “the system” and allowed me to focus on taking notes, linking things and whatnot. It is a breath of fresh air. Properties and queries are amazing. Having youtube embed? So useful.

But I’ve noticed I’m wary of using cmd+z, because I’ve lost blocks and content. I avoid using any special block formatting because using a QUOTE destroyed my cursor position and removed a paragraph. I can no longer edit my custom.css on mobile and I mostly use Logseq on mobile devices.

It is an amazing tool and I love it so much. I’m not paying either so I’m not demanding anything just… I really wish to see Logseq grow and succeed and I feel that as it is, I can’t really recommend 100%.

and_yet_it_moves · May 9, 2023, 2:08am

By coincidence, another issue arose illustrating this very problem. See #7519 and the end of thread Rename query fields when renaming a page title.

And yes, the timing is a pure conincidence - I saw the closing of the issue above only a few hours after having posted the above post.

I’ll quote myself from the github issue:

In the larger picture, I thinkLogSeq has fallen into a trap, that many software projects have seen (not with a nice ending): Core functionality is lacking, therefore the part of the user base consisting of dedicated long-term users is small, the user turn-over is large with many new users coming in but leaving after a short time. Creating new shiny features does bring on new users - but again, they don’t stay long. Large user turn over → many new registered users and Github stars are rising, which is interpreted as a positive sign, when it is actually a sign of a problem. The team thinks all is well and continues creating the new shiny features and ignoring the basic functionality and usability. It’s an evil spiral pointing upwards, but it consists of nothing - it’s like an inflation bubble.

I have seen too many software projects failing because of this effect. It would be a shame if a tool with the potential of LogSeq took this route.

This needs to be fighted somewhat similar to inflation: don’t keep inflating, instead go back to the basic mechanism of the economy/software and do the groundwork to make an actual, stable improvement. The fundamental, unsexy, stabilizing groundworks. To win back a larger base of dedicated, long-term users.

Emil Eifrem, founder and CEO of Neo4j, with Neo4j being somewhat unmatched on the combination of company valuation and user satisfaction the last 10 years, has his “one single advice” on creating a successful software project that comes to mind:

“As a software tool company, you can never, ever, over-index on the satisfaction of your actual, core, daily users”.

Junyi_Du · May 9, 2023, 2:33am

Sorry for the inconvenience.

insufficient bandwidth to maintain the current features

The very fact about the query system is, we already have a devoted developer (Gabriel) on fixing query bugs.
It’s caused by the nature of query DSL’s engineering challenge, as query is a highly user customizable, heavy coupled with other components. We hope some insights on making the query system more maintainable & bug-free. Our current practice is to improving the test case coverage (of course, not only for query, we always adding more tests and introducing more QAs).

Welcome joining the effort on code & unit test case contribution.

weakness of the documentation

We are improving the document & developing a hand book system for on-boarding.
Also community help on improving document would be greatly appreciated:

Junyi_Du · May 9, 2023, 3:07am

And, always, Logseq keep bug fixing at highest priority and stabilizing features is always the development focus.
You may find most code submissions are bug fix & test / QA setups.

francescob · May 12, 2023, 6:30pm

Appreciate the answers!
I’m a paying user, which doesn’t mean anything since Logseq is a free software, but I started paying because of how useful the core functionality is to me.
I started using it at work and I now use it for my life too, it’s my backup memory and instead of having to dig, I find everything.

The new whiteboard addition, while not something I might directly use, seems cool. I could see occasionally using it.

I’m not sold on the AI stuff since I doubt it would be useful for me personally: Logseq is still bad for “long prose”, since everything should be a block, so I’m not sure how valuable AI would be.
That being said, I’ll keep an open mind.

As long as bugs are getting closed over time, I’m happy, in the end the core functionality is very valuable. Everything I’m hoping for is now enhancements to queries/maintaining tags and I’m sure it will come over time.

Also, I really don’t want to migrate to Obsidian where I’d need dozens of plugins to achieve what I do with logseq (tagging + querying)

I’ll keep paying and hopefully my money will allow closing some bugs!

I was thinking if I have to wish for something, it would be Logseq for teams to generate documentation (internal and external). I could see that being a paid feature with the live documentation being hosted.
We use Notion extensively and we just keep losing track of things because the lack of tagging + linked references.

Joe_Smoe · May 12, 2023, 8:31pm

I’m a paid user whose also used the software for the better portion of 2 years now. This thread resonates deeply with me. Reoccurring and obvious bugs that make their way into each release has become a common theme. There’s also core features that are missing some basic functionality to make them complete with no signs they’ll be completed anytime soon.

I’ve been advocating this for some time now but to me their attention went straight to Pro features over a year ago, and with that they’ve become overloaded with too much to do, but that leaves me to question why they would devote energy to the recently added features like number list support, and the addition of the upcoming AI changes tienson has been working on? That time instead could have been focused on improving the core foundation of the tool, which IMO is far more important. I guess it ultimately doesn’t matter what any of us really say though since it is a free tool, and the devs can choose to do what they want.

Precisely.

This is another bug that was introduced in 0.9.5, I guess none of them saw the issue before they pushed out 0.9.6, so its still there in the latest version. It’s an obvious bug too. If you edit ANY blocks anywhere, you could end up with blocks no longer appearing in the reference panel. The only way to fix it? Well you have to reindex, or undo changes to the block. It’s stuff like this that makes me paranoid to trust my notes are safe, and I can’t really recommend Logseq to anybody in its state: Blocks in references panel disappear and remain missing after making changes to the block (Makes references unreliable and unusable in this state) · Issue #9365 · logseq/logseq · GitHub

Video Demonstration:

johnl · May 13, 2023, 5:23am

This is what I was getting at with my Keep focus on augmenting human cognition
I’m sure whiteboards and AI can be useful - but not at the expense of the groundbreaking core features.

Siferiax · May 13, 2023, 5:48am

I must say I agree with many of the sentiments.
Having integrated Logseq deeply into my life in only a few months shows me how valuable this app is for me.
I mainly use the app on my Android phone and am always super weary on upgrading to a new version in case something breaks my workflow.
I always first upgrade my PC and test things before considering my phone.
I went from 0.8.17 to 0.9.5 on my phone due to breaking bugs introduced in the versions between.

I think there’s a balance between new features and bug fixing.
What I’ve mainly been disliking though is the focus on new features that don’t necessarily support the core function of Logseq.
I think there are loads of smaller valuable features to add, instead of the big changes like whiteboards, AI and even a database version.
Logseq can offer a lot of value out of the box and there’s a lot of potential to increase that value with smaller enhancements.
I feel the focus should indeed be on making the core the best it can be first. Through bug fixes, but also enhancements.

I do think there is focus there. But what is currently lacking is communication. I’m still looking forward to more information on the team’s focus, vision and road map.
This will probably also help with perception of the community of what the team is focused on.

For me personally I will continue to use Logseq in my day to day life as it brings me much value as is.

brsma · May 13, 2023, 11:02am

Another (paying/supporting) user here who can only second this concern. I’ve just gone back to Obsidian, as well, due to Logseq’s frustrating increasing unreliability and sluggishness (especially but not only on iOS/iPadOS where it’s barely usable for me). I’ll keep up my financial support for 3 more months in the hope that the development team will sharpens focus, address the quality issues, and stabilise the core product to a point where I can use it professionally again rather than chasing shiny new things without any apparent sense of direction. There is so much potential in the product that I would love to see realised. That requires a better product sense though that would bring coherence to development.

Fingers crossed.

laike9m · May 13, 2023, 5:02pm

Paid user here (I ranked 8th on the all-time donation leaderboard). Big +1, all the replies resonate with me.

Overall, logseq is a lot more stable than it was one or two years ago. OTOH, tiny bugs still appear from time to time.

I do see most commits being bug fixes, which I appreciate. But as other people have mentioned, enhancement of the “core” feature seems stopped or at least not being given a lot of resources. If you look at the feature request list, the highest voted requests have been there for a long time, but no one seems to be addressing them, or at least put them on the roadmap where everyone can see.

That’s not to say that Logseq should not invest in new features. I do think AI writing will be very useful, just as someone else might feel the same for whiteboard. But they should NOT be the only focus.

Final thought: as I said on Twitter, I hope the Logseq core team still has control over what to do, and not the investors:
https://twitter.com/laike9m/status/1657422768407973890

Bader · May 13, 2023, 5:21pm

Hey there,
I’m the contributor responsible for creating the list, but I haven’t received a response to my question yet. It seems that the limited bandwidth made it more of a hindrance than a help. I am no longer able to manage issues.

chessteline · May 13, 2023, 6:12pm

I wish to say that I agree with the general sentiment presented in this post. I use LogSeq on a daily basis and I can’t praise enough its combination of ingenuity and ease of use at the same time. I changed several PKM systems and eventually settled on LogSeq. I don’t want to go any further because LogSeq seems to resonate with the way I take notes and think perfectly well. However, I share beforementioned concerns. Endless stream of annoying regressions with each new release is often frustrating. There are longstanding bugs which are not addressed. Adding new features is cool, but for heavy LogSeq user like me, personally, stability of the core product is the most important feature. I understand that dev team is small and that they have their priorities but I believe that communication between users and developers could be improved on issues that matter the most for majority of users.

overthere · May 13, 2023, 6:48pm

To me it also feels that development is not focused on the core enough. Now, I only test/experiment with LS for now, and check the repo occasionally, so I might be wrong here. Personally I haven’t encountered any serious bugs/deal breaking issues (my limited usage might be the reasons). But still adding things like whiteboards and AI features seems strange for application that doesn’t support some basics like tabs ([Partial Done] Add tabbed windows to desktop client / Multiple instances of Logseq) or even replace (Find and replace a word in multiple occurrences).

As I said I don’t follow the application closely enough to speculate why is that the case, or if it’s reasonable. But I prefer software that focuses on its core first, and adds trinkets only when it is done.

WQing · May 13, 2023, 10:32pm

enhancement of the “core” feature seems stopped

We definitely need more enhancements, but personally, the last several months had brought more enhancements that I consider major and benefited me greatly, than what I got out of year 2022.

one is “enabling backlinks for block references in properties”
the other is “showing page names in default query tables”

both of these I had been asking since 2021 and I am so so happy they got implemented.
(this year also had the query builder, which didn’t benefit me but I think many people appreciated it.)
but if I’m asked what’s the highlight for me in 2022, it would be the enhancement on backlink filtering… and maybe another something, too, that I can’t think of off the top of my head.

so my perception is that the pace of “great enhancements” implementation has been more or less the same, if not faster. — however, I didn’t write this to object your point or defend the team, not at all. rather, I want to use this opportunity to give examples of the kind of enhancements I consider crucial — more than any shiny new features — and would like to see more resources go into. For example the next big enhancement I’m looking forward to is the query builder for advanced queries.

I want to emphasize that the point is polishing (stability) and QoL improvements rather than “community requests”, because a few people (including elsewhere like reddit) have said something like “look at the Feature Requests here, why is the team doing whiteboards and AI instead of these top FR?” which I actually disagree with, because EPUB annotation is just as “new feature” and “big project” as whiteboards and AI; those are things of the same nature. The examples I gave are enhancements and actual pain points. EPUB annotation isn’t. It’s a new feature that has already been handled by other software (not only is calibre also open-source, it allows annotation links that can be very easily dragged into logseq) — the lack of unison of being able to do everything in logseq is “some pain”, but the same can be said for a lack of any non-existing feature, like AI.
Of course, there are many top FR here that are actual enhancements. I just don’t want the team to get the wrong idea from this thread so I’m being pedantic.

anyways, my opinion is this pace of enhancement implementation needs to quicken still, and I agree with all the sentiments in this thread. but I can also understand why the team prioritized what they prioritized:

whiteboards:
- it was part of the founder’s vision from the beginning; so I really can’t blame it. it’s good to have a vision and actually work towards your vision.
- obsidian also released canvas in the same time frame so on hindsight the timing worked out for them
sync and logseq pro: they need to materialize the business model, understandable too
ai: I can imagine they’re under a lot of pressure from the competition; this feature is also quite desired, to be fair
database: also to materialize their business model because they need to get collaboration working

(edited to add)

in-app handbook: people have been complaining about onboarding for a long time
numbered list: well this one i can’t justify (more on this in my post below)

I just hope from now on they will dial back on new things and look back to improve the old.

laike9m · May 14, 2023, 8:12am

Thanks for providing your perspective. I added my reply in a new thread

Junyi_Du · May 14, 2023, 12:17pm

If you look at the feature request list, the highest voted requests have been there for a long time, but no one seems to be addressing them

Basically we have tags on workload estimation / on roadmap, for highly voted tickets:

Junyi_Du · May 14, 2023, 12:54pm

Also, I have an article about Logseq’s practice on QA. I may mention more on our bug-fix priority & issue managing.

Feel free to leave your comments.

Junyi_Du · May 14, 2023, 1:00pm

I just closed some duplicated ones, while leaving those related open

Admittedly, there’s a bandwidth issue - given so many noise and false reports, I may overlook some really important comments. Especially it’s a reply in an old thread. So please ping me (e.g., Discord, as I may have > 100 GitHub notification per day, but I promise I’m trying to check each of them) when there’s something important.

And we need some advises on process enhancement.