Thread by @travisgerke

After 6 years, I have left the tenure track. I’m beyond excited to begin a new #postac role (Director of Health Informatics) next week.

I do have some parting thoughts about the unique situation data scientists face in academia currently. 1/

There is increasing recognition that everyone *needs* a data scientist, and this is often distinct from a statistician (which everyone also needs!!). Data are messy, projects need reports/dashboards, figures should be elegant, reproducible pipelines/data sharing should happen.

Data scientists treat these problems as top priority. Statisticians need all that infrastructure to *just work* because they need to operate at top-of-license conducting inference.

Unfortunately, such data science roles are not often recognized as high value in traditional academic/tenure track paths. This was described incredibly well in a report from the Berkeley Institute for Data Science

https://twitter.com/travisgerke/status/1010177560905420803

@seanjtaylor discussed this with @casualinfer: the reward structure in academia is to publish as many papers as possible, preferentially first/last author to achieve status and promotion.

This contrasts with collaborative non-academic atmospheres, where data scientist teams work to solve current problems at a company.

Though I was successful (FTR I was on track for tenure with 70+ pubs and funding above expectation), the academic reward structure never resonated deeply with me. Like many data scientists, I’m most happy when collaboratively contributing to applied solutions.

Herein lies the academic pitfall for data scientists: the path to promotion/tenure is paved with R01s. Data scientists as PIs of such grants are *exceptionally* rare. Data science problems are not perceived to hold the same priority as more traditional research questions.

Beyond this, the grant-getting game is *slow* and basically a lottery once the application passes a certain quality threshold. I have submitted grants in one cycle to positive reviews with suggestions, then after addressing the suggestions, gotten a worse score on the next cycle.

I can’t shake that this is a painful waste of time. Data scientists thrive on creating efficiency wherever possible; however, as a paradigm of inefficiency, the grant application process can negatively impact your job satisfaction in dramatic ways.

Further, funding uncertainty was already high; it’s only going to get worse with our current national plan of “we have no plan” for safe COVID-19 reopening. With a new leading cause of death for years to come, it’s hard to imagine steady NIH funds for other diseases will persist

All told, I want to be clear that my experience in academia was not bad. I had tremendous mentors, and an institution/dept highly supportive of my collaborative research path. In the end, the path just wasn’t the key to happiness for me for broader infrastructural reasons.

I write all this down because I’ve been approached by data science interested doctoral students/postdocs in recent years with questions about career paths in academia. They exist, but should be pursued with caution.

I'm open to discuss these things by DM any time, feel free to reach out!

Lastly, h/t to a handful of people to follow whose content is useful/inspiring for anyone considering #postac #altac moves @FromPhDtoLife @ProfessorIsIn @cjcornthwaite @clcaterine