Because, even closer to the beginning, in the real first sentence of the article (by which I mean, the first sentence after the title), it says, "This is a mildly edited transcript of my JSConf Australia talk."
This is not an example of the "big text chopped out" paradigm from magazines, it's an example of the, "each slide contains a brief of the most important part of what you'll be saying while that slide is being displayed" paradigm from conference lectures.
For a reader, you get into the rhythm of the writer's voice, only to be interrupted by a giant image of text every other paragraph. It's a battle between just skimming the slides, or trying to read the content.
If the author had just swapped out the "loud" text images for subtle headers, the article would actually flow.
The most annoying thing is I can never be sure whether the BIG TEXT CHOPPED-OUT is actually going to be repeated again later in the article. Otherwise I'd just happily ignore those big sections every single time. As it is now, I have to read it and then get mad when moments later I find myself reading the exact same thing. I wish there was a way to stop this.
edit: Thanks to those commenters who mentioned https://en.wikipedia.org/wiki/Pull_quote - It's nice to be able to properly refer to that which I despise.
This doesn't necessarily justify the first slide's presence at the very beginning, but it might explain it.
I think they're more useful in a newspaper because they can catch the eye as the reader scans the page. On the web they are fairly useless.
Just came here to say you're not alone; I've always hated this too, and I really wish it hadn't gotten adopted on the web
Of course this article format is kinda pushing the limits I agree.
I admit to often reading things that are far outside my wheelhouse, or not always topics I assume I will enjoy. If I start one of these and find myself loosing interest, these quotes and call outs can draw me further in or push me right out.
The image caption, I took it as medium's implementation of although text for cases the images don't load.
In this case they just happened to be similar.
But this article does it at the beginning... we read the sentence "Hello, I used to build very large JavaScript applications" three times. First as an image, then as a label for the image... and then as the first sentence of the article. Why lol