undefined | Better HN

0 pointscolah39y ago0 comments

All we can do is work hard to build academic support:

* In the last three weeks, we've had 80 outreach conversations with various stakeholders for Distill. The majority of these have been academic researchers. The response has been extremely positive.

* A number of ML faculty at Stanford / Berkeley / Toronto / Montreal are very excited and supportive of Distill.

* Distill's steering committee consists of recognized leaders in ML and data visualization.

* We've registered with the library of congres / CrossRef, dotting our "i"s and crossing our "t"s to be a serious journal. In some senses, we're more legitimate than some notable venues.

* The largest industry research groups institutionally support Distill.

My sense is that the academic community really wants to have something like this, if it can be done well. At the end of the day, we need to publish outstanding content and demonstrate that we're a high-quality venue.

0 comments

arjunnarayan9y ago

Can you share a "behind the scenes" of what it took to get Distill off the ground? You hint at dotting your "i"s and crossing your "t"s, but an explicit manual would be useful. Other communities than just machine learning could benefit from something like this, and if Distill succeeds in being taken seriously by your research community, it would help to have a playbook in which to replicate that success in other research communities as well.

caesurae9y ago

My concern is also the academic & industrial support community will support the concentration of a few contributing institutions to such a journal. I have no doubt that Distill will have high-impact and visibility among various audiences.

Yet I don't see how this will readily support possibly cutting-edge work or new research in machine learning that does not have access to visualization development, or these forged connections to Distill to facilitate the development of these visualizations.

So it seems like a likely outcome is that Distill publishes content from well-regarded institutions and increases publicity for that work, to the detriment of a vast bulk of papers which do not have access to the visualization resources to develop Distill-ed versions of their work.

Furthermore, and this is a larger disciplinary issue, but it seems inherently this could end up spotlighting more CS-y machine learning vs statistical learning due to cultural differences between disciplines and differences in computational/web development background in grad students and researchers in both fields. Are there efforts to reach out to statistical associations as well?

j / k navigate · click thread line to collapse