Not sure what I did this past week aside from finishing a post: I read very little.
Well, to be fair on Wednesday we pre-recorded Thursday’s meetup for PyBCN, and on Thursday I moderated the live questions, so there’s that, too.
On Thursday too I tried a newly released Oculus Quest game, Warplanes, available on Sidequest, the alternative/WIP games store. Not sure why it’s not on the official store yet: it is definitely more polished than most, and for me it would be reason enough to buy a Quest. I love flying games, and there are none on the Quest rivaling Warplanes. You play on either side (on launch only Central powers campaign was available, on Saturday the first few planes of the Allies and the campaign were made available). Below you can see some footage of me flying a Fokker Dr.I:
My summary of the Lakehouse paper by Databricks.
This is a very informative talk. Inverted indexes have a large implementation cost, but the results can speak for themselves.
I’m not sure I support this. I’m all in for better sharing of ideas, opinions and dissent, but I prefer the approaches suggested by L. David Marquet (Turn the Ship Around, which I strongly recommend, Leadership is Language which I’m still listening to, and I’m unsure).
So the “hot hand fallacy” from Tversky et al is actually wrong. The reasoning is a bit Monty Hall-esque, but at last the hot hand is back.
An excellent essay on Commoncog.
We bought a bag of MSG a year or so ago, and sprinkle a bit on occasion (on frozen pizzas, or in even home-made dishes). Think of it as “magic salt”. It will make the food tastier in a way. I definitely recommend you add it to your pantry.
That’s an interesting Wikipedia page, I was talking about AUC (area under curve) and ROC (receiver operating caracteristic) with a work mate and ended here.
Not a fan of strict management processes but this looks pretty sane, at least in the way the work is structured.
These are welcome news. It can make a big difference for internal development documents.
Woah, 2021 is starting strong.
This is pretty good news. Looking forward trying Superset again (third time’s the charm?) and seeing if the nagging issues we found are gone.
This is purry interesting
I think the learning as compression concept has a lot of more behind.
SQL on Flink looks good. I have no real time streaming use case right now, but when I do it will be hard to decide among KafkaSQL, Spark Streaming and Flink.