At times it can take a couple of minutes for MyBinder to launch a container, even if the image is already available.
For example, rather than have just the big spinny thing, have a hero image / carousel that could:
- showcase some other Binders, eg recently popular ones ones that are popular over an extended period;
- report on MyBinder stats / costs;
- highlight news: Jupyter releases (and security fixes);
- display Jupyter community announcements and calls (workshops, events)
- highlight recent papers or news stories where Jupyter tools were used.
I appreciate putting this together would detract effort from core service, but it could provide a useful comms channel to the wider community. (In terms of skills requied, it would also open potential for contributory support to web designers / webdevs in putting this together? As the community gets bigger, it needs to call on a wider range of skills to sustain its activities.)
I think this is a great idea!
Variations on it have been discussed a few times but so far no one has gotten around to doing something about it. Ideas that lead to a concrete proposal or PR would be great. I’d help get that merged and deployed.
I need to improve my perfect GIF game but still it amuses me
Additional ideas (that don’t require curation) in no particular order:
- Animated SVG with mouse input
- Sweet JS ‘Game(s)’
- Programming tutorials / fortunes
- Image / data labeling tasks ala ReCaptcha
- Jupyter tips and tricks
- SO issues taggged e.g. ‘Jupyter’ with and without answers
In the interests of demonstrating my personal and individual capacity to credibly curate notebooks well paired with the often highly subjective content contained in BinderHub hosted notebooks, furthering my agenda, and increasing the UX for everyone, I’ve prepared a playlist of videos that I think we can all agree optimally represent the Jupyter and Binder projects at all times:
- “Hall & Oates - You Make My Dreams Come True”
- “Redbone - Come and Get Your Love”
- “Orbital - Halcyon on and On” (Hackers Soundtrack)
- “Eagle Eye Cherry - Save Tonight”
- “Goo Goo Dolls - Name”
- “Dave Matthews - Ants Marching”
- “Fatboy Slim - Praise You”
- “Thirty Seconds To Mars - Up in the Air” (Restart and Run all)
- “Let It Go” (Frozen Soundtrack)
- “Avicii - Waiting For Love”
- “Avicii - Wake Me Up”
- “Symphony of Science - Onward To The Edge”
- “Snow Patrol - The Lightning Strike (What if this storm ends)”
- “Macklemore feat Skylar Grey - Glorious”
- “Dave Matthews Band - You Might Die Trying”
- twitter /westurner/status/1097897764724400129
I may choose to add more videos and/or MIDIs that I like later. A “Keep Listening” option may make JupyterHub GCP instance launches even more desirable.
Fortunes / tips ideas:
- How to add notebook metadata that will be indexed by search engines
Heh heh… BinderRadio…
Reminds me of http://ds106.us/ds106-radio/ …
It also strikes me that there’d be an opportunity to sell sponsorship space… (filthy lucre arguments aside, things like MyBinder do cost someone $$$ to run, although that requires another person on the team - managing ad sales… A restrictive approval policy on what ads are acceptable could also help mean ads were likely to be relevant to the wider community. I’d be wary of going down the adtech / profile your user route tho…).
Does someone know someone at readthedocs? They have been running ads for a while to support their platform and invested a bunch in making their ads not-repulsive. So maybe we can get some advice from them.
The cynic in me thinks that we (mybinder.org) by far don’t get enough traffic (90000 clicks per week) to make a dent in the hosting cost (~$1000 per week). Does anyone know how to estimate ad revenue from number of visitors?
From an UX perspective, a notification telling me my binder awaits me would go a long way – while your ideas are cute 'n all I will still most likely do something else in other tabs.
Do you guys measure the abandon rate, i.e. binder started but never used?
Would be a nice feature to contribute.
mybinder.org does not track what users do inside their sessions. This means from an analytics point of view every launch is a “bounce”.
Re: abandin rates; I read that as the difference between a binder page redirecting to the notebook server (ie user gives and closes tab) or whatever page user wants to open on and the binder launch page being closed before it redirects? Is that easily measured? Is it useful to know?
I don’t know anyone at readthedocs but I do know a thing or two about estimating costs and some about advertising.
At 90k clicks per week, you are WELL within the possibility of advertising being worthwhile. A $10 CPM could pay ~90% of your maintenance costs (where does that budget come from, by the way?). I could build a model to help demonstrate this stuff under uncertainty (computational mathematician here, after all), if you’re interested, but in my opinion all you need to know is that “influencers” with 10-20k engaged followers can rake in thousands a month. It’s all about curating the right relationships with advertising partners.
Kind of related to this discussion: could you have some example binders on the mybinder home page, so if someone arrives there they can click on something and launch an informative example? Or alternatively have a “default” URL so if you click launch without filling any fields it still launches? At the moment it’s not obvious what a new user should do if they’re directed there.
That is a good idea. We could copy some things from http://jupyter.org/try as a start and then replace with new/other/popular repos.
I think a blocker so far with editing the landing page (and other pages on mybinder.org) is that we currently have the design we have, which doesn’t foresee changes. Maybe someone with an eye for design could make a suggestion for small changes to create a new layout that was meant to have these additional elements in them. Compared to us “randomly” adding things. If anyone knows someone like that … please put as in touch