Liberty

https://www.nytimes.com/2018/10/15/technology/mit-college-artificial-intelligence.html

https://www.business-standard.com/article/companies/amazon-bids-400-mn-for-stake-in-spencer-s-deal-stuck-at-valuation-stage-118101300715_1.html

https://www.ft.com/content/9d7c1434-cc50-11e8-b276-b9069bde0956

I'm done with the Michael Lewis book. A surprisingly large fraction of it is about the importance and value of data and data science, both to make our society function better and keep people safe. It was a fun, short one. Recommended. Of course, people who haven't read the book immediately come out to try to discredit it because it might conflict with their ideology. It's all interviews with people who are or were there. Lewis almost does no editorializing, and a very small fraction of it is about Trump, btw. If there are things that make Trump look bad in it, it's because he did the things described, not because of some bias.

I started listening to the audiobook of this. Good so far. Pretty short (about 5 hours). At 2x it's very easy to follow and basically just as long as a long podcast. Lewis is one of the best at documenting the people behind the big events, and this is no exception so far.

Another new Q&A: http://www.csisoftware.com/wp-content/uploads/2018/10/QA-Oct-9-2018-Final.pdf

https://www.bloomberg.com/news/articles/2018-10-12/eddie-lampert-rode-the-worst-trade-of-his-life-all-the-way-down

Breen has shown with Tyco that he knows how to clean up badly management companies and get the value out (divestitures, spin offs, refocusing R&D and capex, etc). It'll be interesting to see what he can do with these assets. The synergies so far are already pretty impressive.

1-hour podcast interview with biogerontologist Aubrey de Grey, mostly about biomedical research into aging: http://mindandmachine.libsyn.com/life-extension-human-longevity-with-dr-aubrey-dr-grey

If I had to guess, I'd say it's because he's writing about mostly events that took place a long time ago, and about two people, one of which is dead. He's usually writing about more current things that he can follow in real time and with people he can talk to (mostly).

People have been predicting mediocre forward returns (the new normal of low returns) since around 2011. I think it's very hard to predict the future, but it makes people feel conservative and like they're being safe when they predict mediocre returns. In reality, who knows what will happen?

It's something that comes up over and over again in different situations, and I think it's based on some human cognitive biases. When benefits are spread out over many kind of invisible people but downsides are concentrated on a few visible people, most people will go with what is visible and easy to create a narrative around. You see this all the time with farmers. Free trade helps everybody who's buying food, but it hurts farmers. It's easy to create a narrative around those poor farmers, but harder to create one around millions of people having access to slightly cheaper/better/more varied food.

Some counterpoints: http://scottgrannis.blogspot.com/2018/10/just-another-panic-attack.html

10.5x EBITDA for a chemical business seems high. Even businessmen in protected niches like paint, trade at 11x and DWDP has a lotmof cyclical components. Peer EMN trades at 8.3x EBITDA and peer BASF at ~7x (—probably a bit high ex their E&P and refinery business). The only business line within DWDP that deserves a higher multiple is Agro, but it seems to be a small part of the overall company. I think you're overlooking the specialty businesses, which will be what remain after they spin off the two other divisions.

TDG buying Esterline Technologies for $4bn: https://www.transdigm.com/investor-relations/news-releases/news-article/?myartid=2371033

I think the denominator is larger for the Newhouse shares owned than the 20 plus million showed on those filings, therefor a smaller overall percentage repurchased. I could be wrong about the way they present it but Newhouse definitely owns more than that. Hmm, I think you''re right, the proxy seems to say that they own 34,778,200 shares. That would make the pace 0.4%, or 4.8% of O/S annually, if they are being proportional in what they sell back to the company.

‪ANH filing showing that $CHTR prob bought back 0.8% of shares outstanding last month (9.6% annual pace): https://www.sec.gov/Archives/edgar/data/914545/000089924318026402/xslF345X03/doc4.xml‬

Model 3 crash NHTSA crash test results: https://www.tesla.com/en_CA/blog/model-3-lowest-probability-injury-any-vehicle-ever-tested-nhtsa

https://arstechnica.com/gadgets/2018/10/intels-new-performance-desktop-lineup-an-overclockable-xeon-9th-gen-core/

DHS release on the alleged hardware hack: https://www.dhs.gov/news/2018/10/06/statement-dhs-press-secretary-recent-media-reports-potential-supply-chain-compromise

The supply chain isn't all in China. Final assembly and some parts are made in China, but a lot of the value of the products is from outside China (most of it, in fact). Lots of hardware from Japan and South-Korea, almost all software and design from the US, services are mostly hosted outside China (except for the China market), etc. I'm not saying it would be easy to do assembly elsewhere than China because of the massive scale of Apple and because there's a lot of know-how in China that hasn't been as developped elsewhere, but to say that it's all in China is misleading. I also think the trade skirmish is mostly by choice at this point. Sure China has been a bad actor and should be pressured and cajoled to do better, but there's no reason except politics why both countries couldn't be in a win-win relationship.

I guess I'm just saying I don't necessarily believed that Cavium is as good as that architecture can be made by the best silicon team on the market with infinite money and access to top fab space, based on what they can already do for mobile, but only an actual attempt by Apple could show us where the chips fall. Also, the Cavium seems to have been designed to be a "data center" and "cloud" processor, so it optimizes for different things than a laptop/workstation CPU would, and might not be representative of what ARM could do if shaped in a different way. I don't think Apple would just stitch together a crapload of tiny cores in a laptop, f.ex., but rather would probably have 4-8 beefy cores designed for single-threaded performance, maybe with some low-power cores to take over when load is low (like in the recent iPhones). I think if they made the switch, it would be to be more in control of the technologies on which they depend (and not be at the mercy of Intel's roadmap and delays and problems, as they've been for a few years). They could probably also recoup some of the margin that goes to Intel since it probably wouldn't cost them as much to develop these as the billions that go to Intel over however many years those architectures last (but that's speculative), and they could differentiate more in non-performance vectors the same way they've done on mobile (incorporate stuff like the secure enclave, neural engine, dedicated co-processors for certain features to save on power, etc). They can already do some of that by including off-CPU silicon like the T2 and whatever ARM iOS-based stuff is running the touchbar in new Macbook Pros, but I think some features would benefit from being on-die close to the CPU (if they want to do more stuff with ML-acceleration). There are many other arguments against the switch, so it's certainly not a forgone conclusion that they will switch, but I think Intel's recent problems probably didn't help them keep the status quo.

What you say is true for the mobile chips. What I'm saying: We haven't seen an ARM chip designed by Apple for laptop/desktops. There's nothing that says that ARM pipelines can't be widened a lot, L2 and L3 caches increased, clock speeds further ramped up, memory buses widened, etc.. They could get to do a lot more per cycle if the die space budget and power budget weren't so constrained by being designed for battery-powered mobile devices. What I'm saying is that even under those huge constraints the Apple-designed SoCs are doing extremely well. So under different contraints, I think they could very well be competitive with x86. But I agree that this would be a difficult transition to me. But it would be easier for Apple than anyone since they control a lot of the stack (hardware, software, app stores, IDEs/SDKs, etc), and because they're one of the only ones with the experience of having gone through one of those transitions recently.

Piece on Amazon's use of AI and machine learning, robots, etc: https://www.fastcompany.com/90246028/how-ai-is-helping-amazon-become-a-trillion-dollar-company

I don't think it's clear that we know that ARM would be slower, since we haven't seen what an ARM chip designed by Apple for laptops and desktops would be like. Their chips in phones and tablets are already competitive with some laptops despite the much lower power draw and smaller thermal headroom. I'd imagine that if they designed a chip to use a lot more power, they could get something pretty competitive. On a javascript benchmark that aims to be pretty real-world, the iPhone XS came on top of an iMac Pro: http://macdailynews.com/2018/09/23/apples-iphone-xs-is-faster-than-an-imac-pro-on-the-speedometer-2-0-javascript-benchmark/ Granted, this might be because the A12 has a lot of L1 cache, but still, I'd like to see what they can do with a desktop/laptop class chip designed in-house. Recent review of the A12: https://www.anandtech.com/show/13392/the-iphone-xs-xs-max-review-unveiling-the-silicon-secrets