OCTO Speaker Series #10: Melody Meckfessel

[idan]

Hello everyone! Welcome to the tenth installment of the OCTO Speaker Series, featuring Melody Meckfessel, CEO of Observable.

Be kind and thoughtful — we take our code of conduct seriously.

[vawogbemi]

What skills make a good data visualization engineer? Is it more software engineering, data engineering or something else?

[hellonearthis]

I think visual arts skills help, knowing about colour and balance help. Also storytelling and the ability to see what the story the data is telling. Computer skills I find the most used is data wrangling, putting the data into a structure that can be used in visualisations, for example taking data in a tabular format (CSV) and putting it into a heretical format (parent child relationships).

[idan]

When I think about Observable, I think the key decision that was made was accepting that there needed to be some breaking changes in order to make better UX possible. The language is javascript, but it's not exactly JavaScript, and there's an excellent notebook by Mike Bostock that explains the breaking changes. One of the things that trip people up about most notebook implementations is the cell execution model, where earlier cells can be run after later cells, and you have to pay attention to that in order to follow the narrative. Observable decided to make a break from that and make it so that all cells affected by a change run, and that makes it much easier to reason about notebook execution

How do you think the notebook UX needs to evolve? Are there more breaking changes like this which will make notebooks accessible for a wider audience?

[gzbakku]

what libs do you guys use which are fast??

[idan]

How should notebooks be versioned? How should they interact with other versioning software like git?

[Wattenberger]

Good data visualization includes both the technical skills to create the viz and the design skills to responsibly represent the data. As we create tools to make data visualizations easier to make, how can we make sure that creators are properly representing data?

[jessmartin]

What would it take for Observable notebooks to be deployed as part of a production environment? Where's the line between Observable and a PaaS, for instance, Heroku?

[jyjster]

How does Observable enable greater autonomy among collaborators?

[mcmcclur]

This was great - thanks!

In the first few minutes, you (Idan) state that you're not exactly sure what to label as the first notebook and that the notebooks you see today seem to be derived from Jupyter. For the record, Stephen Wolfram claims to have invented the first notebook for interacting with software in 1988; one recent reference where he writes this is this article on the current state of notebooks in Mathematica.

Now, Wolfram is not exactly known for graciously acknowledging his predecessors but I think he's got the patents to back up the statement that his company developed a novel notebook. It's also worth mentioning that Fernando Perez (the lead force developer who started Jupyter) cited Mathematica as a specific inspiration in this blog post - as well as Maple and Sage.