Q: With live public notebooks, is there a way to stage edits before making them public?
[...]
For Free users, we also recommend that they fork the notebook (which would be public), make the edits and merge the changes back.
Although the forked notebooks would be public, it would be a temporary notebook which could be deleted once merged.
So Free Individuals just have to... break their live notebooks if they want to edit them?
Like fine, kneecap your free product if you really need to force people to fork over cash. But Observable purports to be a platform that is trying to democratise data analysis and visualisation for the masses. Feels a bit like Wikipedia adding its own paid tier.
I wouldn't put it past them... "Subscribe to Wikipedia Pro today, to see the most up to date articles (the free tier only gets to see 2 year old articles)."
Looks like it has the non-reactive Jupyter style notebook as opposed to the reactive-style Observable one though. I do prefer the latter by a large margin to be honest. Are there any plans to also support reactive notebooks?
I’m working on an open source notebook with built-in support for React and Typescript. It also comes with a built-in reactivity model and evaluate-as-you-type. To be open sourced soon, but preview version already available at https://www.typecell.org
The SaaS crunch continues. I don't have an objective datapoint but over the past few weeks have noticed many of the 'free' apps I use have made significant changes to the free tier and increased pricing.
One way to ameliorate the effects of the "SaaS crunch" (as a user, i.e. potential casualty) is to stop establishing path dependencies on services altogether when a document is sufficient for the problem at hand. In the case here, that isn't as difficult of a mental leap compared to other things, because of what they are from the very start: notebooks.
These are documents; what we're talking about is something that you can prepare on your own machine and then ship off to somewhere else when it's time to share—whether that's by dumping it onto your own Web space, or attaching to an email.
It should have never been the case that e.g. classrooms, etc. were standardizing on a single service provider to handle this problem in the first place.
VC money is drying up, and many of these startups are going to fail because they don't have any sort of foundation as a business. They don't have a product people are willing to pay for.
Part of the reason people don't want to pay is VCs kept funding new startups that give things away. As that stops happening, paid tiers should start to become normal again, and we can all live happily ever after. But it's bad timing for startups on free tiers who are starting to need money right now.
Even if VCs get frustrated, giant companies will still have little difficulty operating free services as a vague loss-leader strategy, like Microsoft with OneNote or whatever.
That stuff will still exist as competition for any startup.
Yes especially the web version is terrible. I have to refresh it every day and it can't even search across all notebooks.
Of course there's no desktop version for Linux or BSD so I'm stuck with the web one.
The only reason I use it is it comes for 'free' with Office 365.
I wish there was something open source like it. Most open source notebook options are markdown and I don't like that (because it's not wysiwyg) and electron.
All the markdown ones I've seen have a two-pane setup with the raw code and the rendered result. I don't like that at all. It wastes screen space.
Also, I'm not very familiar with markdown, it seems to be really popular with developers. I guessbecause GitHub uses it for readme pages. But I have no attachment to it at all. It feels like an extra hurdle to learn.
I really liked the old tomboy but the development stalled and the later reincarnations didn't have the snappiness of the original. And not having pictures is annoying.
Yeah, this seems dumb or at least short sighted. If you make it free for individuals, but limit collaboration, then you get those users arguing for their companies to buy the service, which is where the real money is.
Let it happen. It’s time for all the unsustainable growth oriented models to stop burning unlimited money and actually charge users what it costs to provide.
Yes, this was a big change and quite a shock for anyone expecting to be able to use the docs to learn d3 for use in traditional web apps, the way they used to use the d3 docs.
Personally I think what happened is that the DOM-manipulating and animation parts of d3 (but not the utility libraries) became existentially threatened by the rise of the component-based reactive JS frameworks, and the d3 authors made the reasonable decision to accept that, explicitly target a data science and academic audience, and create a for-profit company aligned with it. Maintaining docs targeting both use cases would have been expensive, and would have threatened the quality of the docs associated with their primary focus (observable), so they made the decision to accept that the docs would now be quite awkward for the traditional web app use case.
Been a while since I’ve touched observable (not a criticism, just that I’m out of the loop) and didn’t know or forgot they had free tier private notebooks.
Doesn’t seem to me to be a big deal and a relatively normal way to differentiate free and paid accounts for such a service/SaaS.
Anyone have insights on the broader implications though? Business is struggling? Are they being outcompeted by other “notebook” offerings? I’d guess their integration with cloud services would be subpar given that they themselves, last I heard, ran on Heroku and so may never have seen cloud integration as a priority. Is the visualisation and JS focus not paying off in data analytics or data science at least for enterprise? It wouldn’t surprise me (though I have no idea) that they got VC with big goals but would have been better off as a small boutique company.
> Doesn’t seem to me to be a big deal and a relatively normal way to differentiate free and paid accounts for such a service/SaaS.
I'm not steeped in the Observable ecosystem either, but it's worth thinking clearly about what "public" means to not conflate things that are not by their nature inherently joined. If a notebook has to be public because your account is on the free tier, then that alone is maybe not so bad. But what Observable seems to have adopted, just like many others, is a forced social layer indexed by actor on top of their hosted notebook editor/depository.
For a notebook to be "public", by reasonable inference that involves the ability for anyone who comes across it to be able to view it (and in the case here, probably easily fork it). So far, no big deal. Observable, however, just like GitHub until fairly recently, indexes all of your work and then makes that public, too. That's what not having access to private notebooks really means—not being able exercise control over the social aggregator to stop it from putting things in the public graph it makes available to everyone whether you asked to do that or not. It's a little bit creepy how we've just taken this as a given without really noticing the distinction.
The example I always use to make the distinction clear is that when I'm at the post office or the grocery store or wherever, then I'm in public. Those are public places; if you see me around, fine. But what's not happening is that there's not an omnipresent observer following me around to keep a record: "First they went to the grocery store, and then they went to the post office, and then they went back home and stayed there for 6 hours, and then[...]" And it doesn't then take those records and casually publish it for anyone who asks for it. Because that would be fucking creepy, right?
1. You can turn off Location History if you want. (It should perhaps not be turned on by default, but the option to turn it off is there, at least.)
2. The data that is collected by Android devices is private—Google isn't actually making it available to "anyone who asks". In that regard it's a personal aid akin to your browser history and rather unlike your profile on social networking sites like GitHub—who only (frighteningly recently) just began to allow you to change your privacy settings to prevent them from publishing a public index of your activity across their site.
I have to think through how this will impact my usage of observable when teaching. I'm not sure how I feel about asking my students to utilize it if part of using it is that all of their notebooks are instantly public, or if that will run afoul of university-level policies about disclosing class rosters etc.
I suppose that is exactly the reason that they offer educational programs, but for my class size and the resources available for my classes I'm not sure that is available to me.
I am sympathetic. From someone who completed his degree in the last decade, I offer my perspective on this sort of thing.
When I completed my degree and got the eventual one year look back survey, I actually called out the explosion in faculty electing to use third party services which weren’t vetted by or supported by the university. One of my chief complaints was that students had no recourse other than to agree to arbitrary terms and sometimes pay arbitrary amounts to companies that were clearly monetizing our data.
(Although that last part isn’t as much a concern here.)
I can imagine if I was required to use something that made my coursework public, I would have simply gone to the CISO Office or the Student Ombudsman if not offered an alternative.
I'll be honest and say that I - not even that long ago - held similar attitudes and slowly made a series of compromises (to myself) that led to this point. While I'm not sure I fully regret it I do think it was ... not as well-considered as I'd like. ("I don't even know who you are anymore!" he said into the mirror.)
We mostly use a locally-hosted jupyterhub, but for JS work often used other environments. For JS, we made the switch to observable from iodide (after it was wound down) and I think I will re-evaluate platforms again, including JS kernels in Jupyter.
Thank you for having the self-awareness to look back from where you are and recognize where you came from, along with the humility of character to acknowledge that errors may have been made-all while having a sense of humor about it too!
Repl.it is being used at my college for the programming languages. It's nice to share code and not have to setup a programming environment, but I'll admit to at some point simply viewing the forks of an assignment as all the forks are public, and copying some of their code to get my assignment to work.
The usernames don't have to respond to a specific user on repl.it so I guess there's no violations as long as students don't use their real names.
I mean, I guess I should have been more obviously self-aware in my original post, but yes, my intent in that comment was explicitly to bemoan that I was not able to take the deal they offer for education and that my precarious freeloading led to a change in plans.
I "released" a bunch of my class projects as open source and nearly ran afoul of plagiarism accusations (of my own stuff) even though at the time (pre-GitHub) my stuff was only in darcs repos on my own hosts and was generally hard to find. (That ultimately was what saved me in that it was very obvious these were indeed my own files and also that I wasn't posting them for other people/classmates to cheat off me but for my own usage and/or sometimes team collaboration.)
I realize there's a lot of interesting ethics boundaries on public software projects during education. I've always been "pro-open science" that more student projects should be open source/allowed to be open source as "teaching tools in the round", but am sympathetic also to how easily that can be abused for cheating or plagiarizing. I think it is an interesting discussion with no easy answers.
I was expecting this move, since what seemed like plan A (i.e. get companies to pay for team collaboration features) was quite a tough sell in my experience. I've been using Observable for work for quite a while, but generally my colleagues have not been interested in collaborating on any of my notebooks - they tend to consume them, not produce them.
In the meantime the completely free model for using it seemed overly generous given that. Hence we've reached plan B.
I was hoping for at least a limited number of private notebooks, or at least unlisted notebooks, since that would make it a bit easier to use for prototyping private stuff...
A few interesting reactive notebooks off the top of my head:
- Pluto.jl (Julia)
- Clerk (Clojure)
- Polynote (Scala, Python?)
- Percival (Datalog + JavaScript)
As for Observable, note the runtime is open source. I don't know any perfectly polished alternatives that use it, but
for now there's Starboard.gg (which if I'm not mistaken uses Observable's runtime, but someone else said it doesn't), and this VSCode extension: https://marketplace.visualstudio.com/items?itemName=GordonSm...
I'd love to see more alternatives in this space, specifically of reactive notebooks or similar environments.
I have tried out Clerk yesterday and was really satisfied with the plug & play setup, the simplicity of the approach (no need to change anything in source files, just add markdown in comments) and all the features (Hiccup, TailwindCSS, LaTeX, data vis., etc.) it provides out of the box.
But of course it is not a social/collaborative platform like Observable (though I believe you can host your compiled notebooks on your own server). But the developer of Clerk and some others also created a platform for Clojure-based reactive notebooks that seems especially great for learners and teaching: https://www.maria.cloud It has a picture language, evaluation of sub-forms, basic structural editing and is really beginner-friendly.
So this means that drafts will be immediately public or is it just notebooks that you publish “privately” through invitation, a shared link or something like that? (seems more like the former to me, but I haven’t been able to confirm) [EDIT: it is confirmed that there will be no private drafts: https://talk.observablehq.com/t/new-policy/7098/2 ]
In that case, I will not be able to use Observable any longer. I am not ready to have my half-baked ideas, ongoing research and sketches floating around in public space; this would be very confusing to anyone on my profile and some of them contain private data I don’t want to share. But I also cannot afford to pay for another subscription, there is just so much these days that keeps adding up.
Observable has been a great experience, especially when I was focussing more on JS and data visualization and I enjoyed doing explorative work on the platform. I don’t like that privacy and “local-first” these days are considered features instead of defaults. There are many other things that provide enough value for people to pay for and I am very disappointed that Observable decided to force free users into sharing everything.
49 comments
[ 2.9 ms ] story [ 112 ms ] threadLike fine, kneecap your free product if you really need to force people to fork over cash. But Observable purports to be a platform that is trying to democratise data analysis and visualisation for the masses. Feels a bit like Wikipedia adding its own paid tier.
It is for profit
I wouldn't put it past them... "Subscribe to Wikipedia Pro today, to see the most up to date articles (the free tier only gets to see 2 year old articles)."
Wikipedia does have a paid tier! It's a way that very large companies which consume Wikipedia data can query it more efficiently: https://enterprise.wikimedia.com https://www.theverge.com/2022/6/22/23178245/google-paying-wi...
https://starboard.gg/gz
https://github.com/gzuidhof/starboard-notebook
https://github.com/gzuidhof/starboard-observable
More expensive credit means less free offers attempting to lure customers in.
These are documents; what we're talking about is something that you can prepare on your own machine and then ship off to somewhere else when it's time to share—whether that's by dumping it onto your own Web space, or attaching to an email.
It should have never been the case that e.g. classrooms, etc. were standardizing on a single service provider to handle this problem in the first place.
this resonated with me deeply. use =/= pay for, and that distinction wasn't clear for me at first until I had my first taste of saas ownership.
That stuff will still exist as competition for any startup.
Although having used OneNote quite a lot, I think free is still too high a price to pay for it.
Of course there's no desktop version for Linux or BSD so I'm stuck with the web one.
The only reason I use it is it comes for 'free' with Office 365.
I wish there was something open source like it. Most open source notebook options are markdown and I don't like that (because it's not wysiwyg) and electron.
Also, I'm not very familiar with markdown, it seems to be really popular with developers. I guessbecause GitHub uses it for readme pages. But I have no attachment to it at all. It feels like an extra hurdle to learn.
I really liked the old tomboy but the development stalled and the later reincarnations didn't have the snappiness of the original. And not having pictures is annoying.
It's free, so I can't complain.
Personally I think what happened is that the DOM-manipulating and animation parts of d3 (but not the utility libraries) became existentially threatened by the rise of the component-based reactive JS frameworks, and the d3 authors made the reasonable decision to accept that, explicitly target a data science and academic audience, and create a for-profit company aligned with it. Maintaining docs targeting both use cases would have been expensive, and would have threatened the quality of the docs associated with their primary focus (observable), so they made the decision to accept that the docs would now be quite awkward for the traditional web app use case.
Doesn’t seem to me to be a big deal and a relatively normal way to differentiate free and paid accounts for such a service/SaaS.
Anyone have insights on the broader implications though? Business is struggling? Are they being outcompeted by other “notebook” offerings? I’d guess their integration with cloud services would be subpar given that they themselves, last I heard, ran on Heroku and so may never have seen cloud integration as a priority. Is the visualisation and JS focus not paying off in data analytics or data science at least for enterprise? It wouldn’t surprise me (though I have no idea) that they got VC with big goals but would have been better off as a small boutique company.
I'm not steeped in the Observable ecosystem either, but it's worth thinking clearly about what "public" means to not conflate things that are not by their nature inherently joined. If a notebook has to be public because your account is on the free tier, then that alone is maybe not so bad. But what Observable seems to have adopted, just like many others, is a forced social layer indexed by actor on top of their hosted notebook editor/depository.
For a notebook to be "public", by reasonable inference that involves the ability for anyone who comes across it to be able to view it (and in the case here, probably easily fork it). So far, no big deal. Observable, however, just like GitHub until fairly recently, indexes all of your work and then makes that public, too. That's what not having access to private notebooks really means—not being able exercise control over the social aggregator to stop it from putting things in the public graph it makes available to everyone whether you asked to do that or not. It's a little bit creepy how we've just taken this as a given without really noticing the distinction.
The example I always use to make the distinction clear is that when I'm at the post office or the grocery store or wherever, then I'm in public. Those are public places; if you see me around, fine. But what's not happening is that there's not an omnipresent observer following me around to keep a record: "First they went to the grocery store, and then they went to the post office, and then they went back home and stayed there for 6 hours, and then[...]" And it doesn't then take those records and casually publish it for anyone who asks for it. Because that would be fucking creepy, right?
See <https://www.colbyrussell.com/2019/02/15/what-happened-in-jan...>
Or you’d be Google Maps…
1. You can turn off Location History if you want. (It should perhaps not be turned on by default, but the option to turn it off is there, at least.)
2. The data that is collected by Android devices is private—Google isn't actually making it available to "anyone who asks". In that regard it's a personal aid akin to your browser history and rather unlike your profile on social networking sites like GitHub—who only (frighteningly recently) just began to allow you to change your privacy settings to prevent them from publishing a public index of your activity across their site.
I suppose that is exactly the reason that they offer educational programs, but for my class size and the resources available for my classes I'm not sure that is available to me.
When I completed my degree and got the eventual one year look back survey, I actually called out the explosion in faculty electing to use third party services which weren’t vetted by or supported by the university. One of my chief complaints was that students had no recourse other than to agree to arbitrary terms and sometimes pay arbitrary amounts to companies that were clearly monetizing our data.
(Although that last part isn’t as much a concern here.)
I can imagine if I was required to use something that made my coursework public, I would have simply gone to the CISO Office or the Student Ombudsman if not offered an alternative.
We mostly use a locally-hosted jupyterhub, but for JS work often used other environments. For JS, we made the switch to observable from iodide (after it was wound down) and I think I will re-evaluate platforms again, including JS kernels in Jupyter.
The usernames don't have to respond to a specific user on repl.it so I guess there's no violations as long as students don't use their real names.
If the terms of that program aren’t workable, they can lose you. But freeloading is almost by definition precarious business.
I realize there's a lot of interesting ethics boundaries on public software projects during education. I've always been "pro-open science" that more student projects should be open source/allowed to be open source as "teaching tools in the round", but am sympathetic also to how easily that can be abused for cheating or plagiarizing. I think it is an interesting discussion with no easy answers.
In the meantime the completely free model for using it seemed overly generous given that. Hence we've reached plan B.
I was hoping for at least a limited number of private notebooks, or at least unlisted notebooks, since that would make it a bit easier to use for prototyping private stuff...
- Pluto.jl (Julia)
- Clerk (Clojure)
- Polynote (Scala, Python?)
- Percival (Datalog + JavaScript)
As for Observable, note the runtime is open source. I don't know any perfectly polished alternatives that use it, but for now there's Starboard.gg (which if I'm not mistaken uses Observable's runtime, but someone else said it doesn't), and this VSCode extension: https://marketplace.visualstudio.com/items?itemName=GordonSm...
I'd love to see more alternatives in this space, specifically of reactive notebooks or similar environments.
But of course it is not a social/collaborative platform like Observable (though I believe you can host your compiled notebooks on your own server). But the developer of Clerk and some others also created a platform for Clojure-based reactive notebooks that seems especially great for learners and teaching: https://www.maria.cloud It has a picture language, evaluation of sub-forms, basic structural editing and is really beginner-friendly.
In that case, I will not be able to use Observable any longer. I am not ready to have my half-baked ideas, ongoing research and sketches floating around in public space; this would be very confusing to anyone on my profile and some of them contain private data I don’t want to share. But I also cannot afford to pay for another subscription, there is just so much these days that keeps adding up.
Observable has been a great experience, especially when I was focussing more on JS and data visualization and I enjoyed doing explorative work on the platform. I don’t like that privacy and “local-first” these days are considered features instead of defaults. There are many other things that provide enough value for people to pay for and I am very disappointed that Observable decided to force free users into sharing everything.