blevin

January in MT is no joke. Were the activities in that part of the year more on the theory end? Did you have to wait long between your reservation and getting a spot?

We talk about the time-value of money; perhaps there’s also a time-value of time.

These resemble portfolio hedging bets on potential future paths of technology playforms that could affect their ad biz. Google seems to buy a spread of small bets with several-year expiry and high convexity (i.e. big payout if equilibrium shifts in that direction). I suspect a number of these come out of discussions anticipating what could be next after mobile.

A good pod episode on this:

https://peterattiamd.com/irenedavis/

One layer of the internet is about packets, another is about ordered sequential streams — and both are widely supported protocols. Above these could potentially sit a dissemination layer, retaining and persisting anything of interest. But in a legal system where information can be speech or property, it is problematic to do that as a standard protocol. Instead we have archive.org, The Archive Team, BitTorrent, and a million progressively less durable options from there, including some of us running youtube-dl on videos of interest, grabbing PDF’s off Arxiv, SSRN, etc. I’m not sure what to do about any of this except a yearly donation to archive.org, and accepting the great distributed fitness function of fate.

This is what HN was like in the early years -- a (much) smaller community of people making things, sharing them, and rooting each other on.

A high profile example is the Donlin project in Alaska. They've been spending decades on de-risking and logistics, and once the gold becomes valuable enough to extract it's likely to be one of the largest mines in the world.

https://www.novagold.com/properties/donlin_gold/overview/

The lofty ask price may be unduly anchored by the hypothetical value it could have represented to Google in the scenario where something like this turned out to be the next big platform. Web and later mobile represented big shifts of economic activity, and the incumbents all know they need to place a big bet on candidates for any next potential shift, hence many of them making similar moves on VR, AR, self driving cars, etc. If they didn’t, and missed the next big step entirely, they would be judged to be incompetent; if they did and it simply didn’t pan out, no big deal since all the other players were making similar bets. In this context ML may see itself as still representing that potential future value, while others have updated their pricing based on limited uptake of VR and AR so far. It’s unclear where they could find a buyer at that price.

You know how a lot of engineering comes down to making good choices about trade offs? Could be speed vs memory, could be which language or API to use, could be keeping it simple vs anticipating likely future needs. Same thing goes for working with an organization or codebase that has sub-optimal aspects. You think about and pick which, among many axes, is the most impactful thing to push on and you just do it. If the software is huge and poorly documented, write the docs you want to see for the parts you work on. If it’s a big tangled mess, pick somewhere to start unwinding and just find a way to make some sort of progress. If you do this, and you have a healthy org, it will become a natural reference point for others about how to do things well.

If you are instead saying that you feel no hope that work will be valued or recognized, you need to decide either to do it anyway (because you are someone who’s going to think and do the right thing, ie, lead) given the imperfect reality we live in, or you can decide to find a healthier org to work in. By healthy, I just mean more functional and well organized, and thriving in a field of sufficient resources — similar to what you’d think of in a healthy biological organism.

So you might feel constrained by the existing language, build systems, release process, etc compared to working fully autonomously, on the other hand you can accept them as they are, gently advocate for ways they could be improved, but spend most of you energy on the things you can directly affect. It’s like gradient descent — push every axis towards improvement but push hardest on things you can improve the most. If you do this over time and don’t silo away too much, you will naturally gain more influence over time (again assuming a reasonably healthy org).

Eizo sells monitors for color sensitive work which contain built in calibration probes that pop out of a hidden compartment. They can be set to recalibrate on a schedule. They are pretty neat.

https://www.eizo.com/products/coloredge/cg279x/

Did the diffuse reflection reconstruction rely on high speed photography? Or a special geometric setup of the reflector? Trying to think of how this could work.

Did you find the work to be intellectually satisfying and creative, or was it somewhat rote in the sense of having to spend lots of energy on data hygiene?

The arithmetic vs. geometric mean distinction is a good one to note. Since you mention risk-adjusted return, do you favor any particular approaches to optimization? ReSolve makes a pretty strong case for numerical optimization, summarized by this decision tree based on prior beliefs:

https://twitter.com/gestaltu/status/1044977487556595714

You wonder what analysis went into that deal. Just the month before, Uber had announced a $5.2 billion loss for the quarter, which gives a sense of the competitive landscape SuperShuttle faced.

There are also arguments that PE has been doing a lot of poor deals lately and institutions allocating to them are going to be disappointed:

https://americanaffairsjournal.org/2018/02/private-equity-ov...

US RDA is 0.8 grams/day per kg of bodyweight. For a 150 pound / 68 kg person that’s around 55 grams.

The line about the “Apple-designed rubber dome” sticks out. Why even say that? Did Apple not design the other bits? My sympathy for whomever over there is trying to navigate the whole keyboard situation and messaging.

It is the craziest thing, isn’t it, that programs are meant to be both read by humans and executed by machines which operate at significantly different time scales than us?

That’s a great point about the term “intuitive”. Our intuition is based on our own experience, so it is necessarily different for each of us. But we can at least unite around shared experience, our cultural heritage and language. User interface idioms (double tap to zoom), videogame tropes (there is a button you hold to run), or typical standard library calls (here is how to fmt a string) all correspond to communities that know them. The beauty of learning lisp, forth, ML, prolog, verilog is not only that you grow your intuition but also that they help you better reflect on the boundaries of your knowledge and its finite extent. So much disagreement online is simply because we come from non-overlapping communities that use the same words to describe different bound experiences.

This graph comparing 20 year total returns suggests another reason gold is considered by some to be a worthwhile “investment”:

https://twitter.com/charliebilello/status/116655914552655052...

Regarding precedent: the Bank of Japan began QE in 2001 and has expanded it beyond sovereign bonds to also buy equities and ETF’s. Could the day come when the Fed bids up $BYND?

My experience has likewise been that film DP’s have extremely impressive visual acuity, memory for color, etc. Talented artists often have developed whole sets of skills that the rest of us are unaware are even skills. In the same way a programmer might have thought about cache line false sharing as it affects memory hierarchy throughput, visual art often hides lots of expertise you cannot directly perceive, even as you can sense the quality of the whole.

A price inflation measure like CPI is a summary statistic, so it does not capture the underlying dispersion of costs as various people experience them. It is true that imported, mass-manufactured goods remain cheap, such as TV’s. At the same time, domestic resources like housing, health care, and education are difficult for many to afford. It turns out those things are vital to a good life, more so than TV’s.

A monetary supply inflation measure like M1 shows a remarkable rise since 2008. The central bank added lots of money by financing the US government debt, and that has indirectly bid up financial assets, VC, etc. If you owned US financial assets over the last decade you are flush; if you didn’t you are wondering why important things in life are so expensive.

Barclays Global Agg bond index includes $13T USD of negative yield bonds. And Swiss bonds have negative yields out to 50 years.

The basic attention token (https://basicattentiontoken.org/) seems to have the right characteristics and incentives to fill this role. Any ideas why the others have failed so far? It’s hard to insert yourself as a value-adding middleman in an existing marketplace unless consensus agrees the current configuration is in crisis. It’d be interesting if anyone could find a way to do this for Amazon’s commingled counterfeit inventory, for example.

A fun variation is pausing to floss halfway through brushing your teeth — foamy mouth and all. A major function of toothpaste is that it breaks up bacterial biofilm, so it provides a nice buff to the mechanical action of flossing.

The most interesting and least elaborated detail is the Afterburner FPGA. I’m curious what the programming model is (Metal compute kernels?) and how it compares to GPU.

There’s also no natural law that says it needs to be stationary. The historical mean can just gradually rise. It can be like climate change. This is what makes markets fascinating — many independent agents evaluating the present opportunity set, acting on their internalized narratives and beliefs as well as data, with gradual turnover of the population of participants.

In a thread asking for options for present passive income (ie money coming in every month that you can eat with while you learn Python) it seems important to distinguish that need vs a long term investment strategy where you have money passively going out every month instead of coming in. One thing the market provides is a trade between people who have needs on different time frames.

Keep in mind that SPY yield today is 1.9% — below CPI — so you are losing value even before paying tax on that income. If you are hoping for capital gains / equity risk premium to provide extra return, you owe yourself to read a bit about where we are in the economic cycle, current corporate debt levels compared to history, and to look at what happened in 2000, 2008, and last December and visualize going through that. To get passive income beyond the dividend yield you’ll need to be selling shares, not buying them. Consider the scenario, hopefully unlikely but historically demonstrated, that it goes down 50% at some point, or spends an entire decade with zero total return. US equities have been a powerhouse but it’s worth some careful research if you are looking for passive income.

It’s almost as though you have to treat them like human drivers. We cannot formally verify 16 year olds, either, or sufficiently introspect accurate reasons for their behavior. Instead, we require them to pass a test, we apply actuarial cost models, etc.

Makes sense; thanks. I'm out of my depth but it seems neurosurgery may just have it easier here, being able to fix a rigid stereotaxy head frame and fiducial markers across both imaging and therapy. Not to mention less tissue deformation enabling a gamma knife intersection-of-beams approach (i.e. ~200 collimated, mm-wide gamma sources).

Not to be glib but on behalf of the thousands of people going into a radiotherapy clinic today for treatment, thanks for working to improve these techniques.