Yup, bi-weekly Thursday posts are for paid subscribers and are more half-baked short bits about whatever’s on Randy’s mind at the time.
I realized something as I was writing up the in-my-head version of my PyData talk for the post that came out on Tuesday. It was a hair over 2300 words and I think it was a decent approximation of whatever it is I said live on stage. I also realized that given my typical post length on here, I’ve been mass producing 15-30 minute length talks for a few years now. That’s nuts.
The guidelines I had been given was to use 30 minutes for the main talk, leaving 10 minutes for Q&A. The actual talk lasted roughly 25 minutes or so from start to finish and spanned 47 slides in total. By any reasonable measure it was a failure of time management on my part. I definitely could’ve found some entertaining ways to use up an extra 5 minutes.
But here’s the thing. The whole time I was writing the talk, I had absolutely no idea how long the stupid thing would take. I just wrote and planned out narrative lines in my head until completely through vibes only I decided “this should be close enough”. You’d think that someone that’s both obsessed about measurement and time would have a better method, but nope.
For example, there doesn’t really exist a “speaking time formula” like you might expect in a world where we have readability formulae. At best you can find some very rough guides about how many people speak at about 150 English words per minute in short bursts, but over long stretches it averages out to closer to 100 wpm.
If you peek at the Wikipedia page for “Speech Tempo”, all you wind up finding is a short page explaining how measuring “how fast people talk” is fraught with complicated definitions. The definition of “word” across languages is troublesome so people tried to measure things via syllables per minute. But languages also have different levels of complexity for their syllables, so even that confounds understanding how fast people can speak. There’s also issues with how the same speaker can exhibit a wide variation of tempo to adapt to different situations like telling a story or giving directions.
So yeah, none of that is particularly helpful in understanding just how to fill up a certain amount of talk time.
If we skip the interesting academic question of measuring speaking rates for ALL human speech and just focus on English, things seem only slightly clearer. There’s posts by vendors and such that list out metrics on speaking time, showing that depending on the situation people vary between 100 and maybe 200 wpm for most use cases. Beyond that, you’re just gonna have to figure it out for yourself.
Now I’m sure you’re thinking, “Randy, quit whining and just GIVE THE TALK.” A few practice runs would almost definitely give a good approximation of how long a talk would go. But, unlike some rare folks, I don’t write up a whole talk in full, commit it to memory, and then deliver it. My memory just isn’t that good. Heck, Tuesday’s post exists because I was re-enacting the thing in my head, not the other way around.
Instead I just make sure I know my basic points really well, and use the slides as guideposts as I ad-lib commentary along the way. As I’m writing the talk, I’m reciting ideas for what I want to say to the slide as I go. My sense is that I don’t repeat myself very often each time I go around. If I come across something I want to repeat, I have to work it into the slide somehow as a reminder. This of course leads to a lot of variation in between deliveries. It’s great for adapting to the reactions of an audience, but garbage for time budgeting.
Anyways. I’d love to find some better way to plan out a talk to fit an allotted time slot better. I had originally hoped that I could ad-lib enough gap filler to make things work out, but somehow the momentum of the talk somehow worked against me and kept pushing me along. How the heck do other people do this anyways? They are most certainly not planning every last word out in advance. If anyone knows I’d love for them to tell me their secrets…