Intro

Swapcard is an app used at EAG(x) conferences for keeping track of the agenda and scheduling 1:1s. Swapcard is less than ideal for some purposes, so EAG(x) organisers usually share a spreadsheet with attendee data with participants.

I looked at few hundreds profiles for recent EAGx and here is what I learned:

• Profile writing tips
• Mass reading process
• Swapcard woes

Profile writing tips

• fill it out sooner rather than later, this way more people can see up-to-date data
• use Job title and Company to express meaningful information, even if it is not a perfect fit for your situation
  • these fields are (the only ones except name and photo) visible on the attendees view
• put more specific things on your profile, e.g. some sample topics in your area of expertise/interest that you are happy to talk about
• summarise external links (e.g. to bio on another website) so the content is indexable
  • unless your goal is to make them less searchable, then put details you want less discoverable behind a link, e.g. to a google doc
• maybe avoid putting references to specific EAG(x), otherwise if data is not updated for new EAG(x) profile feels abandoned
• controversial: if emojis fit your style, they can make for nice variation in paragraphs of text
  • 🤖🧠🧟‍♀️🦐🐁🐔☄️🧬

Mass reading process

Caveats

Spreadsheet data is not perfect. Some data can be missing, some can be not up to date.
This is probably only worth it if you want to look at hundred(s) of profiles, it requires upfront time costs.

Process

1. Download the spreadsheet (File → Download → Comma Separated Values (.csv))
2. Programmatically extract profiles of interest by specific criteria
   • e.g. people who have expertise in "AI safety technical research" _but not_ in "AI strategy & policy"
   • or people who work for specific company
3. Sort profiles by specific criteria
   • e.g. length of raw text fields — if there is more information, it is easier to decide if you'd be interested in reaching out to that person
4. Convert profile data into more readable representation
   • e.g. change fields order/drop some fields

Now you can review profiles most relevant to you with the most important information a glance!

Implementation

I used python via PyCharm to read CSV, filter, and convert to Markdown, here is exactly how. Then I used Obsidian to review profiles.

LLMs are great at writing filtering functions, once you describe data and your criteria, and tweaking markdown generation.

Swapcard woes

• Swapcard is only available online, no cache in the app
  • it would be nice to work on setting up meetings on a flight
• infinite scroll for attendees
  • hard to know how many people fit your search criteria
• not clear how to mark people you've seen
  • can one do better than filter by first/last name and remember letter where you stopped?
• filtering criteria allow only _or_
• limited sorting options (default, first/last name, registration date)
• requires clicks to see longer text fields