Knit picks: May 14

Hi Knit newcomers! We have a few new friends this week because of

Newsletter stuff

Substack has been backfilled with old Knit picks! Some of the more popular issues are linked from the about page.

The Substack cleanup lines up with sending a link out to the Flatiron alum Slack. I may try to expand the newsletter circulation more in the future. The exposure is daunting but I hope to encourage participation.

Web stuff

I very quickly wrote a data flow builder GUI. Even in its extremely rough state, it shows the data diff feature more quickly than I could before. I plan to use the builder GUI to prototype interactions, and maybe even to get some user feedback.

Gretzky stuff

The Great One is known for skating to where the puck is going to be. If you don’t anticipate you will always be behind. And projecting into the future lets you spot openings that aren’t obvious in the present.

So it’s always been perplexing to me that ambitious entrepreneurs use traditional TAM/SAM/SOM market sizing. They assume the current market size and project how much their business can capture as it grows. But markets are changing too!

If I had a tweed jacket and an armchair I would propose a new measure, Potential Future Market (PFM). Let me know where I can pick up my Nobel.

Predicting stuff is hard though! Thinking about users is easier. So instead, consider who is the most underserved user? Paul Graham says these folks have “hair on fire problems”¹. In more subtle cases, underserved users may not realize it, like when everyone was trying to get the latest Razr while Apple was out inventing the iPhone.

So who are the most underserved data users, who are also likely to be an emerging market?

My current hypothesis is:

• Data system builders. The folks who build data platforms. Inside of companies, these are data infrastructure engineers, who take off-the-shelf tools and figure out how to configure and customize them for a specific organization. They can also make their own products to sell as a data product vendor. I would argue tools for building data tools suck.
• Open data collaborators. I often point to the JHU Covid-19 GitHub, which has 17,000 forks but only 8 committers. Clearly there is interest to contribute or improve the data, but all the current tools fail. Data vendors are so busy capturing value from enterprises that they almost entirely ignore non-enterprise markets. There is no equivalent of open source for data².

In the eternal words of Mean Girls, “Gretchen, stop trying to make PFM happen!”

PS. A note to new and old friends that I am very receptive to feedback and questions. Just reply to this e-mail!