15 Comments

Japan Business Insights covers innovative smaller cap Japanese tech stocks. Have taken a pause of late in publishing new pieces, but a small library of past write-ups can be found here: https://japanbi.substack.com/

Expand full comment

Really interesting stuff. You asked about newsletters covering particular industries; We just launched ours , covering political stocks. Check it out: fineprintdata.com

Expand full comment

Fun fraud on an indian ecommerce platform Meesho (it is not the usual ecommerce platform) but one for India2 or emerging lower middle class internet users - gets resellers to sell goods to their friends - think Amway or Tupperware on steroids - https://entrackr.com/2021/07/the-mystery-of-meesho-ghost-orders/

Expand full comment
Jul 17, 2021Liked by Byrne Hobart

I'm glad I just read that Eric Nelson interview. I think he has a few blind spots but I had never before spent even thirty seconds thinking about the relationship between conservatives and the dead-tree publishing industry, and it is interesting.

Expand full comment

Training large neural networks is important. However it is slow and expensive, and de facto requires large clusters of NVIDIA GPUs or Google TPUs with very fast interconnect. End to end back propagation via gradient descent does not distribute well.

What if we could train our models using lots of cheap commodity CPUs with slow interconnect? That would be truly wonderful! If worked well, it would ruin a big chunk of NVIDIAs server GPU business. However I have been working on this for over 3 years, and have not yet succeeded. EMH says that it will not work.

https://www.isaacleonard.com/ml/distributed/

Expand full comment

https://sci-hubtw.hkvisa.net/10.1002/wcc.82 A nice article on the history of climate engineering. From 2010.

Expand full comment

“ Getting more data does not mean getting better answers. “ -> made me laugh, because of the futility in telling people this. This is especially true in my day job, VC / Growth. People find comfort in numbers, when in fact the numbers are distracting you from what truly matters. Much of what works in the world is due to qualitative bets made, and if you are right, then the quantitative stuff falls into place based on how correct your qualitative bet was. Some many people think like engineers in venture, and I’m sure there are people for whom that works, but most people could get a process done in 50% of the time with all the information needed to make as good of a decision.

Also, as an ex-oil/gas infra investor, I love seeing the Prize there - along with what I said above here, if you ever have a friend who says they can forecast oil prices, have them read that book. :D

Expand full comment

Anatomy of the Swipe by Siddiqui (2020) was a great read — the kind of book I wish existed for every industry/vertical

Expand full comment