The mistake here is sizing. Pabrai put 77% into $MU, which sounds like conviction but is exactly what guarantees you cannot hold it. At 77%, every wobble in the stock is a wobble in your whole net worth, so you find a reason to sell, right before the move you were right about.
At 10% sizing, none of that happens. A drawdown is a few percent of the book, easy to sit through. And if the stock does 15x, that 10% position grows into 150% of your original portfolio, so you more than double your entire book on one name without ever being forced out.
Conviction is not the position size. Conviction is staying in the seat.
With all the forced pseudoscience projects with questionable outcomes emerging from Indian Institutes of Technology across India, degrading decades of scientific credibility, mostly from political and religio-cultural agenda from the BJP and the Ministry of Ayush, this is a timely piece. Never before seen kind of "secret war" for those who really love this nation. Science community in India needs to be stronger than ever and come together to fight this spreading blight and growing rot. The hapless public deserve our candles in the dark.
If GLM5.2 or DeepSeek V4 is as good as Opus or even just Sonnet (personally for coding, with search tools provided, I feel it easily matches Sonnet/5.4) then you realise what it means?
More layoffs in tech
Only friction to Claudemaxxing is API costs right now