tdullien

13 ms·

121.
by tdullien 3y ago
I graduated from high school in 2000, and sold my first company to Google in 2011, insisting on cash and not stock given the performance of the market through my entire adult life ;)
122.
by tdullien 3y ago
Imagine you take all sorts of copyrighted pictures, compress them lossily, encrypt them separately using a unique passphrase that can be guessed, and place them into a big tarball to be distributed. Are you infringing copyright? I think thi
123.
by tdullien 3y ago
I haven't checked in depth, but ... while the Fed funds rate has oscillated a bit ("noise") the curve for 10- or 30-year t-bills or mortgages is much smoother, and newly issued junk bonds will track those more than the Fed fu
124.
by tdullien 3y ago
Doesn't that have more to do with Michael Milken pioneering junk bonds, which then vastly expanded the available credit? As in: The fact that the 80s had comparatively high interest rates doesn't matter to the argument, as the nec
125.
by tdullien 3y ago
"it depends" - but there are several components at play: PE benefited greatly from a long-term decline in interest rates. The amount of debt a company can service at a given profitability is directly related to the current prevail
126.
by tdullien 3y ago
Xoogler here. GCP was not an abstraction on Borg when I was there. GKE isn't either. So up until late 2018 when I left, very little of Cloud ran on "proper" Google3 infra. This may have shifted slightly (cloud has been fishin
127.
by tdullien 3y ago
My parents (west German) drove Ladas, first one bought in the late 70s (which was actually reasonably solid, built under license from Fiat. Came with a large very solid cast iron set of tools, and a big pot of paint in the right color). Ful
128.
by tdullien 3y ago
50% of marriages fail, 25% of surviving marriages are unhappy, so the odds of marrying, not divorcing, and being happy are a bit above 38%. 70% of divorces are initiated by women, and more in higher income brackets, yet only 50% of men unde
129.
by tdullien 3y ago
Go's GC is much more naive than modern Java GCs, and the codegen is not terrific. That said: A few design decisions in the Java language (that are only slowly getting fixed) harm performance: 1. "Everything is virtually dispatched
130.
by tdullien 3y ago
Any insights into what went wrong?
131.
by tdullien 3y ago
Isn't that unusual for normal venture backed companies? I know preferred shares with high dividends as debt-like instruments for financial companies, but ... are there venture investments that expect dividend payments?
132.
by tdullien 3y ago
I support this.
133.
by tdullien 3y ago
Oh wow, my toot on HN, rated higher than any of the cool things I have done :-)
134.
by tdullien 3y ago
Yep
135.
by tdullien 3y ago
The big mistake Megaupload made was charging copyright holders a fee to delete a link to their copyrighted content, but not deleting the content on the backend. This was what allowed the prosecutors to bring out racketeering charges, brin
136.
by tdullien 3y ago
Wait, I thought piecewise-linear interpolation search was well-known, and known to be similarly performant to the Google "learnt indices" paper? This paper seems to imply that previous approaches didn't use piecewise-linear a
137.
by tdullien 3y ago
Or are prices going down because in a high-interest scenario dividend yields have to go up? ;) Causality in econ is hard .
138.
by tdullien 3y ago
The fact that software falls under copyright was a conscious decision in 1978 because they couldn't find a better place to put it under; so "writing software" was equated to writing books or poetry. The point here is that cop
139.
by tdullien 3y ago
I am still absolutely baffled that people think weights are copyrightable and hence licensable. There is no reason to believe they are, which means any restriction placed on the weights themselves is bullshit.
140.
by tdullien 3y ago
When channelling Oppenheimer, it is worth remembering that von Neumann quipped: "Some people profess guilt to claim credit for sin."
141.
by tdullien 3y ago
Are you sure you're right about those nines?
142.
by tdullien 3y ago
TF in its first version was stellarly misdesigned. It was infuriatingly difficult to use, particularly if you were of the "I just want to write code and have it autodiffed + SGDed" school, I found it crazy to use Python to manuall
143.
by tdullien 4y ago
We decided that animals can't create copyrightable works and hence limited the ability to create copyrightable works to humans . I am fine with granting AIs the ability to create copyrightable works provided we grant that right, and
144.
by tdullien 4y ago
So there is a strong argument that the construct the Fed chose for the SVB depositor bailout was very much designed to give them free hand to continue rate hikes.
145.
Importing 4B chess games using ES and Universal Profiling(elastic.co)
2 points by tdullien 4y ago|0 comments
146.
by tdullien 4y ago
Iirc it's based on crosvm, so AWS didn't do much "make it open source" as "adapted an existing piece of open source".
147.
by tdullien 4y ago
I had the same impression g . Any suggestion on where to look?
148.
by tdullien 4y ago
Yeah. I really like the ideas proposed by Brendan Gregg -- essentially encouraging every HLL runtime to embed an eBPF-based unwinder in it's own executable. The upshot of that would be "generic, in-production unwinding of native c
149.
by tdullien 4y ago
What everybody in this discussion seems to miss is that you don't need to unwind the DWARF data structures during profiling time , you are free to convert DWARF to a fast-lookup data structure on the machine. DWARF needs to support ev
150.
by tdullien 4y ago
For prodfiler.com we convert .eh_frame (DWARF) unwinding format to something that is more similar to ORC and optimized for lookup in eBPF data structures.

More ›