Thanks all for the replies. [This reply turned into a ramble over a coffee break! Not sure what point I was trying to make in the end... Apols for that...]
Re: trying to log confusion, I'll try to keep notes on things I notice or that crop up in student / novice comments. We're about to do a trawl of comments on a MOOC that used the notebooks so I'll try to tease out any of those that mention notebook usability issues. I guess one of the immediate complexity issues relates to things like keyboard shortcuts. Eg when a single notebook cell could only be moved up and down using a toolbar up/down button, you were quite limited in commands available for rearranging cells. As the ability to select and move multiple cells was introduced, more really useful functionality became available, but at a cost to novice users: there are lots of different ways in which you can move cells around rather than just one. The possibilities become bewildering. More generally, give a novice an instrument with too many dials and buttons on it and it can be hard to get them to see which are the useful primitive operations and which are the nice to haves that allow you to work more powerfully, or do some fine tuning. (I'm all for getting novices to play with power tools, but think there's often a danger of overwhelming them by making too many features available too early.) In an apprentice model, you earn the right to access more skills/knowledge/tools (eg in something like Stack Overflow, features are unlocked as you gain points). Same with growing with a tool that is being actively developed - you start to desire additional features to help you work round problems you've identified with the current one, and as each new feature is developed and introduced it is essentially 'unlocked' and added as a more specialist tool the tools you already use. I wonder what an interface would look like that reveals features to you the more you use it? (Not just ungreying things previously locked; absenting them from completely from the UI in the early phases of use). So everyone starts with their own "version 1" when they come in and the functionality of version 2, version 3, version 4 etc is revealed as they become more expert... (The tutorial helps here, but I can imagine the tutorial getting richer. I'm also really wary of this being seen as some sort of reimagining of clippy (young 'uns won't get that reference!), that pop up unannounced every so often to show you a particular feature!) Introducing new features into a familiar interface also brings with it surprise, and as you realise you can now do something you couldn't before you essentially give yourself a little lesson in how the new feature works and make a mental note you can now do this thing that you couldn't before. I can see the JupyterLab approach working in education at a more macro scale. Eg as you progress through several courses, you might be given a JupyterLab config with more things enabled by default at the start of each more advanced course. (Developers do this with their own tools all the time by installing addons and plugins; but they build the complexity / power of the environment as they grow more skilled in using it.). --tony On Thursday, 16 February 2017 22:34:08 UTC, ellisonbg wrote: > > Tony, > > Great questions and feedback! Inline thoughts below... > > > - will Jupyter notebooks continue to be available as such once > Jupyterlab is > > available? (could a jupyterlab instance be configured to just mimic a > simple > > notebook UI, for example) > > Yes, the core JS component in JupyterLab are very well decoupled and > designed to be used and reused in different ways. The JupyterLab team > is very aware of the need to provide a UI that remains as simple as > possible. Allowing individual notebooks to be edited in their own > browser tab is *one* way of providing that. We have other ideas as > well and will continue to explore these things for the 1.0 release and > beyond. I spend much of my time teaching undergrads with the notebook > and I know first hand how important it is to have something simple. > > > - will Jupyterhub continue to supplort multiuser deployment of Jupyter > > notebooks? > > Yep, already works fine with JupyterHub. I have been using > JupyterLab+Hub for the past 2 months in my data science course. Simply > pip installing jupyterlab (and enabling its serverextension) was all > that was needed. > > > > > One of the attractions of Jupyter notebooks in education and for > supporting > > use of code outside computing discipline is that the notebook interface > is > > relatively simple and friendly without all the sidebar chrome and menus > and > > features and tools and stuff that make IDEs a terrifying experience for > most > > people. I'm keen to be able to keep using a simple clean interface from > the > > off with students and nontechies. > > Yes absolutely. Keeping JupyterLab simple while still providing the > extra power has been one of my main design priorities. We are *not* > there yet, but this is super important. > > > > > I'm lobbying my institution to make notebooks available locally, a > process > > that will take 3-6 months to get deployed and then be expected to not > change > > much for a chunk of time. What should I be lobbying for?! I'm pitching > for > > novices to be able to access simple notebook UIs without any need for > too > > many features in the first instance. > > It depends on when you are doing this. > > If it was today, I would just stick with the classic jupyter notebook > + jupyterhub. When JupyterLab comes out *midyear* (I don't know where > Matthias got the idea it might be delayed until 2018 - that isn't the > case), it should be easy to optionally add JupyterLab (literally just > an extra pip install), but to keep classic notebook as well. Both run > fine side by side in the same JupyterHub deployment and a user can > pick which to use if both are installed. > > > > > (eg if/when nteract ships with a bundled kernel, I can see it being > great > > for use with this user community (even more so if the ability to launch > > temporary or multiuser nteract instances from a remote instituional > server > > accessed via a browser).) > > Yeah the most difficult thing to install is not the UI - but the kernels. > > > > > By way of trying to express my general concerns, as opposed to just > being > > critical of the new... Looking at things like RStudio, it used to be > > relatively simple... but as it gets richer features and more powerful > tools, > > and experts who've grown with it just have to keep up with it, and maybe > > also demand more of it as they get more expert/professional, it just > gets > > more complicated and scary for novices coming to it for the first time. > > Sometimes less is more. The on-ramp needs to be kept simple. (Or at > least, > > it helps if there is a simple on-ramp somewhere...). There are risks to > > always developing more and moving away from the simple offering that > > appealed to folk in the first place... Which isn't to say I don't like > the > > new features that appear with each new release of eg Jupyter notebooks > and > > isn't to say that Jupyter project shouldn't become an ever more powerful > > tool for expert and power scientific computing users. > > +100. > > > > > But it's easy to forget that the experience of welcoming new features as > you > > perceive they're missing, because you're growing in expertise as the > tool > > acquires more powerful features, is different to someone coming to the > > environment for the first time - as a simple notebook that did what it > did 3 > > years ago, to the more complex notebook it is now, to the yet more > complex > > Jupyterlab view? > > Can you be more specific about the "complexity" here. I am not denying > the notebook has become more complex or that JupyterLab is more > complex than the notebook. But there are many types of complexities > and I am interested about which you are thinking about here. Some that > I can think of that could be an issue (in any software): > > * Multiple "acitvities" on a single page > * Complexity of the code base > * Visual complexity > * Feature complexity (too many visual elements, too many borders, > regions, colors, etc.) > * Bad UI/UX design (distracting, overwhelming, etc.) > > Would love any specific concrete feedback you have on these things to > help us refine JupyterLab. > > Cheers, > > Brian > > > > > --tony > > > > On Wednesday, 15 February 2017 15:11:30 UTC, takowl wrote: > >> > >> Hi all, > >> > >> I've just released the first beta for Jupyter Notebook 5.0. Please try > it > >> out and let us know about any bugs. You can install it using: > >> > >> pip install --upgrade --pre notebook > >> > >> There isn't a headline big new feature in 5.0, but rather a range of > new > >> features and improvements. You can read about some of them here: > >> http://jupyter-notebook.readthedocs.io/en/latest/changelog.html > >> > >> Thanks, > >> Thomas > > > > -- > > You received this message because you are subscribed to the Google > Groups > > "Project Jupyter" group. > > To unsubscribe from this group and stop receiving emails from it, send > an > > email to [email protected] <javascript:>. > > To post to this group, send email to [email protected] > <javascript:>. > > To view this discussion on the web visit > > > https://groups.google.com/d/msgid/jupyter/3d9410e4-096f-42a8-b5a7-43a88b5635d6%40googlegroups.com. > > > > > > For more options, visit https://groups.google.com/d/optout. > > > > -- > Brian E. Granger > Associate Professor of Physics and Data Science > Cal Poly State University, San Luis Obispo > @ellisonbg on Twitter and GitHub > [email protected] <javascript:> and [email protected] <javascript:> > -- You received this message because you are subscribed to the Google Groups "Project Jupyter" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/jupyter/e9b222e0-e568-4287-a101-8b1b3b0e1579%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
