anon1253

There is a good map on the wiki of Concentrated Solar [1] showing the Direct Normal Irradiation. The US West coast is in somewhat of a unique position (as is Australia, South Africa, and a couple of others I suppose): high density population centers in areas with a lot of solar radiation. In those areas it might very well be the best method available (dunno, could be). Transporting electricity at distance is non-trivial however. Plans to put these installations in the Sahara and transport it to Europe via high-voltage DC lines have pretty much gone nowhere (https://en.wikipedia.org/wiki/Desertec). So non-competitive was more a reference to the constraints that it imposes, but I could've been clearer for sure.

[1]: https://upload.wikimedia.org/wikipedia/commons/c/c7/Global_M...

Straight up dumping it in the ocean is generally considered bad. The standard trick is to wash/mix the brine with fresh ocean water to reduce the concentration of salt before dumping it back. Another interesting byproduct that is often overlooked are the metals and minerals. Things like lithium (and even gold) are relatively abundant in sea water and could be recovered from the brine (at the cost of complexity of course).

Absolutely, but then again (almost) everything is a fragile ecosystem. And if man-made climate change is real (I'm inclined to believe that), then it's less of an absolute and more of a trade-off ... a tricky one certainly, but already we're seeing deserts expand (https://en.wikipedia.org/wiki/Desertification) and there are efforts to prevent it (https://en.wikipedia.org/wiki/Great_Green_Wall) so it could potentially be another tool.

Yeah it does limit it a bit, and if the goal is really to plant trees, then eventually it will be somewhat self defeating as forests cool the air and will generate clouds. On the other hand, desalination doesn't require extremely high temperatures, so it also kinda depends on the heat capacity of the molten salt. It probably also doesn't really have to be 24/7/365. But that's I guess a numbers game, so obviously it has to be modeled first.

Yep! but then with fewer greenhouses so it has more of a https://en.wikipedia.org/wiki/Desert_greening effect

These types of systems are generally not competitive with PV for electricity, since they require a lot of concentrated solar energy thus limiting the options for deployment. However, they are great for generating heat. And with heat you can trivially desalinate water. Since trees can be a nice carbon sink, I've often wondered if this type of scheme might make sense:

- Plant heliostat molten salt reactors in desert coastal areas (the concentrated solar kind)

- Use generated heat to desalinate water (e.g. Multi-stage flash distillation)

- Use desalinated water to provide irrigation

- Plant (potentially GMO'ed) trees and other bootstrapping organisms to "green" the desert

- Rinse repeat.

The net effect is that you can potentially turn large swaths of desert into forests, with minimal external energy requirements (and potentially build an agricultural community around them, more or less self contained in terms of water/energy)

There is no need to train this particular model, but adopting this (or any novel model, the field is moving fast) to say Dutch, Italian, Hungarian, Icelandic or whatever still requires training. Luckily for most languages they are provided (at least in the case of BERT, FastText, or regular skipgram). But there is also still quite a bit of leeway in domain specific adoption (for example SciBERT for scientific texts, or legal and financial documents) reddit/wikipedia does carry a bias. Each of which not only requires pretraining the model, but also generating a huge and fairly well formatted corpus. And, although the parameters are usually finetunable, it does break down sometimes on the various sub-word tokenizations used.

"The cost of training the model from scratch using our code is about $50k."

Still a substantially steep curve for a bootstrapping startup. It's something I continually run into myself. I have somewhat of a weekend project trying to build a search engine but man ... the cost of just the SSDs and GPUs is daunting on a regular salary. As the complexity of these models grows, so does the barrier to entry for a regular joe like me; which is a shame I think. I know in the US it's fairly normal for a data scientist to pull 100k+ / year, but in the Netherlands salaries pretty much stall at 40k (and angel investment in IT/AI is at an all time low). More generally I fear this will become a bit of a sociotechnical issue if complex AI models will be out of reach for entire economies (especially for cases like language because not everyone speaks English and "minor" languages like those in EU countries are a massive market to explore, yet hard to get into).

I found the readme of https://github.com/tdebatty/java-string-similarity particularly helpful, and of course the (Java) lib itself. It's not comprehensive by any means, but it gives a good starting point

I tend to work a lot in Lisp and XML, both are more or less trees if you squint (with the Lisp syntax famously being the AST due to homoiconicity) and it always makes me wonder if there are better command line tree search or tree diff algorithms out there (extra awesome if it works with git merge strategies). I mean whitespace preference is fine and all, but sometimes you just don’t care :p

Knowing some C/C++ is very useful if you want to get a better understanding of what the computer is doing. Or Assembly of course, but that gets complicated fast.

Common Lisp / Clojure / Scheme (or any language from that family) will really let you appreciate some fundamental functional programming tricks once you finally "grok" them.

Prolog, it's great. You'll never look at graphs, trees, search or first order logic the same way. And, maybe, some day you'll be the one that says "that's just two lines of prolog".

Haskell, or any other very strongly statically typed language. I'm definitely partial to dynamically typed languages, but you gotta try it before you knock it :-)

Also if you have some money to spend, try Mathematica. It's one of those things that keeps getting reinvented, but poorly. Shamefully it's not open.

Regex named capture groups are somewhat of an underused feature I find, especially with things like simple parsers they help to keep stuff readable https://regular-expressions.mobi/named.html

There are several metrics in addition to the k-annonimity (https://en.wikipedia.org/wiki/K-anonymity) that has been going around the past few days. For example l-diversity (https://en.wikipedia.org/wiki/L-diversity) and t-closeness (https://en.wikipedia.org/wiki/T-closeness)

t-closeness, is defined as: An equivalence class is said to have t-closeness if the distance between the distribution of a sensitive attribute in this class and the distribution of the attribute in the whole table is no more than a threshold. A table is said to have t-closeness if all equivalence classes have t-closeness.

In short: the distribution of a particular sensitive value should not be further away than a distance t from the overall distribution.

Using the t-closeness metric circumvents issues associated with k-anonymity and ℓ-diversity. Briefly, k-anonymity states that a certain attribute class should be present in at least k records, which introduces ambiguity in the data set. However, if each of the k equivalence classes are the same, properties could still be resolved simply by elimination. The ℓ-diversity metric circumvents this problem by adding a further requirement: in addition to the class to being seen in k records, these records must have at least ℓ ‘well represented’ values. But if an attacker knows the real-world distribution of values, then attributes could still be disclosed with a certain probability, simply by combining different data sources

Strength exercise seems to help https://www.webmd.com/men/features/exercise-and-testosterone...

Ever since I was a child I had this strange idea that I was missing fingers (I have 10 like most). From grade school till middle school I even kept counting them as if something was amiss. I mean, it's not like it's actively bothering me or something, but just a bit of a curious quirk: I think I should have 12 fingers. Although in my case it feels like they should be to the side of my pinky, not between the index and thumb. As far as I know my parents didn't have them removed at birth either. Funny enough I believe the gene for it is even dominant (meaning it needs only one copy to express the phenotype), it's just not very common in the general populous.

"The datastream will include a specially designed primer, or set of simple scientific principles and data that would be common to all extraterrestrial intelligences, providing a common base of reference to enable those receiving the signal to commence the mammoth task of decoding the encyclopedia. /The message will be accompanied by a short video message by Wikipedia co-founder Jimmy Wales, and images required for the re-creation of fundraiser banners/" (emphasis mine)

Seems like they're very serious about their banners, glad that gift to the world has a chance of living on in the lives of extraterrestrials in the case of our inevitable demise!

For a second I confused this with TerraGen (https://planetside.co.uk/) and got all excited about reliving a nostalgic youth.

It's also actively harming medical and scientific progress as I allude to here in this talk https://www.youtube.com/watch?v=EM61rn9Gxl4&list=PLjzcwcP9P2...

My go-to model that might work is:

- Plant heliostat molten salt reactors in desert coastal areas (the concentrated solar kind)

- Use generated heat to desalinate water

- Use desalinated water to provide irrigation

- Plant (potentially GMO'ed) bootstrapping organisms to create a soil (greening)

- Plant trees

- Rinse repeat.

A lot of "blast from the past" comments! But as a counter example, we still use Jena extensively (in combination with its server Fuseki) to deal with biomedical ontologies and taxonomies for use in Natural Language Processing (e.g. Human Phenotype Ontology, Gene Ontology, etc). We even recently made the step to add some PROLOG-ish inference rules [1]. I have nothing but love for the RDF ecosystem, in the sense that I love their ideas even if some of the implementations are a bit wonky. For example, the performance of property paths (i.e. the + and * operators in sparql) leave things to be desired sometimes. Not to mention the funny looks from some devs when you say you use RDF, but I take that as a badge of honor! It took me a while to get it, so I finally decided to write down what it all means a couple of years ago [2].

[1]: https://jena.apache.org/documentation/inference/index.html

[2]: https://joelkuiper.eu/semantic-web

My favorite term is "annoying sized" data, not enough to warrant clusters and HPC, but enough to make a decent laptop crawl to a halt. It's that uncomfortable in-between that makes up the bulk of the data I usually encounter.

As an amateur astrophotographer I’ve been following the development of cubesats closely. It seems more and more likely that a small group or an individual would be able to launch a space based telescope within my life time. Right now it’s a bit tricky still, and not really sure if you’re allowed to have a camera in space yet (might be considered espionage). But still, extremely cool ... maybe someday soon a crowd funded mini Hubble might be an option!

Love it. Especially the echo state network trick. I wonder how much of BERT/ELMO performance is simply due to them having a such a high dimensionality. Not that there is anything wrong with that, just makes a tad less practical for some applications.

Yep. We've tried to run compute intensive tasks on the cloud but it was a total disaster. With 24U worth of Supermicro we're able to do it with ease with a 10x reduction in costs. The cloud has its place, but nothing beats bare metal for speed (and in some cases, reliability)

This is true for lower apertures, but at higher apertures they become much more noticeable especially at high magnification. This is especially annoying for telescopes (it tends to look like this https://farm9.staticflickr.com/8679/16584213212_5b33222d6e_c...), which why astrophotographers use a trick called flat calibration to remove any artifacts from dust or smudges in the optical train. The trick is surprisingly simple: take a picture of an evenly illuminated surface (usually a flat box or a clear sky) and make sure your ADU is in the linear phase of the sensor (ergo, don't clip the highlights or darks). Then divide your original image by the flat ($T * mean(flat) / flat). This is of course easier with a monochrome sensor and works best with RAW data, but it can definitely be done with a color CMOS as well. In case you ever find yourself with hard to correct dust on the sensor: this trick might save an otherwise poor image.

+/- 2 (https://en.wikipedia.org/wiki/The_Magical_Number_Seven,_Plus...)

Okay, I’m going to be a little bit far out there. A lot of this seems to be grounded in the negative. “How to deal with x” how to “do y when z has failed”. Some of it technical and some of it interpersonal. Why have we created this sphere around programming? What makes /this/ profession so different from any other? And my believe is: nothing. It’s not hard. It’s not better. It won’t “eat the world”. Heck it probably didn’t make that much of a difference in the grand scheme. It’s just sitting on your ass and either do as you’re told or come up with something better, like 90% of all work. Software is just like any other job, and chances are it is completely stupid like most other desk jobs out there. And if not, the same general skills apply. Stop feeling like it is difficult or even special.

As an extreme tangent, but something that I was reminded of due to "tubes".

I recently bought my first audio upgrade (Cambridge DAC 100, Little Dot MK III with Voshkod 6ZH1P-EV from the 70's on Beyerdynamic DT 1990 Pro headphones) and honestly, the first time I listened to it I was in tears. Coming from just plugging in some Sennheiser HD1 to my mobo audio out, the difference was staggering. There were sounds in tracks I was quite familiar with that I had never heard before. I bought a Schiit with a closed DT 1770 Pro for my office ... which wasn't nearly as good (and incidentally I think the filter caps might not handle the el cheapo DC-AC converter on the solar panels all that well which gives annoying (as in really audible) cracking and popping sounds even after tube rolling). Now for the downside: "audiophile" is a horrible niche to be in, there is so much cargo-culting and downright hostility that makes it a very "toxic" place to be in (not to mention stupidly expensive if you buy all the ahum "bs" ... a signal processing class would be a better investment...). But, I find the toxicity to be a little bit sad as well. Since it detracts people from a very real life improvement. Audio is one of the primary modalities of our sensory system, and even with mediocre hearing (like a male 30+ will likely have) you can, and do, hear an enjoyable difference for a modest investment. The difference between some second-grade Bluetooth thing and a nice "desktop" set /is/ real. My advice, get a nice open headphone (400-600$ range, I /really/ recommend something with a detachable cable since those fail first), something to replace your on board DAC for 150$, and an headphone amp (I like tubes, but for no other reason than that I like the way they look and "sound"). For something around 750-1000$ your life will be better and it'll last you a heck of a lot longer than any AirPod will (heirloom piece). That is, if you like listening to music, if not, disregard everything. Yes it's an investment, but this is HN, where I feel recommending 1k is not that big of a deal.

Now for the more on topic part: apart from some high voltage DC applications like transceivers or transmitters (like microwaves!) tubes have largely been replaced by solid state. And with good reason, they last much longer, and are less conductive to all sort of ailments ... so apart from /super/ specialized applications they have no place. Apart from maybe providing that nice warm glow on your desk, when you're lonely, at night, and just want to listen to that "one song" while staring out of the window.

I really like this idea of composing selectable attributes as schemas (ala RDF). I've been fairly reluctant in adopting specs into my own code, mostly because in a lot of cases I just didn't see the benefit (not for the things I typically do, not saying I wasn't interested in generative testing and the likes ... those still seem very promising but also hard to get right). Especially, when dealing with "distributed" things (like the world wide web, where everything is maybe and only some things useful). And, RDF got this surprisingly right. I was sort-of half expecting him to also introduce PROLOG into it, but maybe next time. The idea of collections of subject-predicate-object statements forming an isomorphism with graphs, and sets of these statements form concrete "aggregates" plays extremely well with the idea of unification through search.

Somewhat tangentially, it also keeps reminding me somehow of Alfred North Whitehead's concepts of reality as process flows https://en.wikipedia.org/wiki/Alfred_North_Whitehead#Whitehe...

"In summary, Whitehead rejects the idea of separate and unchanging bits of matter as the most basic building blocks of reality, in favor of the idea of reality as interrelated events in process. He conceives of reality as composed of processes of dynamic "becoming" rather than static "being", emphasizing that all physical things change and evolve, and that changeless "essences" such as matter are mere abstractions from the interrelated events that are the final real things that make up the world."

'The lot is marked “VORHES”, though some articles call it “Voorhees” or “Voorhis”'

I think this might be a derivation of the Dutch word "voorhuis" (https://translate.google.com/#auto/en/voorhuis). The transliteration is "front house" (as in front office, portal, hall) and usually denotes a street level entrance to a larger complex. Given that New York had a lot of immigration from the Netherlands (heck, it was called New Amsterdam for a while) I think this might make sense.