electrograv

You're right, I initially completely misinterpreted your point -- sorry about that. I edited my response accordingly.

My new point is simply this: You can make this argument about anything whose value declines relative to other things; it doesn't make it a useful interpretation of the actual market trends.

For example, I could say that the price of computational resources hasn't been falling over the past few decades due to advances in computer efficiency an speed; no, rather I claim that this is just an illusion -- the cost of computation has actually remained fixed, while the price of literally everything else in the world has been rising, as long as we denominate them in terms of computational resources! But this is silly and doesn't reflect what's actually happening, because I'm subtly making an axiomatic/unfalsifiable assertion here (that the 'true cost' of computational resources is fixed).

It might be easy to misconstrue as reflective of truth because many things do rise in cost, so it intuitively might seem like a profound alternate viewpoint on reality that logically checks out. But, in reality, the viewpoint is simply wrong / not useful -- a reflection merely of an axiomatic assertion yielding a tautological claim, with no external evidence to back it up.

Edit: My original response totally misunderstood your post, below is an updated response:

When Bitcoin's fiat-denominated value drops, the claim that the price of everything actually increased while Bitcoin remains stable — becomes a clearly silly argument when you realize we can make the exact same argument for literally any other asset of your choice (which will in some cases yield completely contradictory conclusions about whether the price of everything else is going up, or down).

To make this argument, you are essentially axiomatically proclaiming that the value of X is fixed, and so when its value appears to drop relative to other things, this therefore is actually just a case of everything else getting more expensive.

The problem is you actually have to justify why you believe the value of X is fixed; otherwise, the conclusion is tautological. Without justification, you will find endless logical contradictions by applying the same logic simultaneously to multiple assets, yielding stark disagreement/contradiction on whether "everything else" went up or down in value, depending on which asset you chose as your denomination reference point :)

If you re-read my posts, you'll note several things I point out or at least allude to, beyond the oversimplified characterization of "not subjective":

(1) We have an objective science that disambiguates and understands differences that are subjective versus universal preferences in speakers (to some reasonable threshold of population universality; e.g. maybe you find some 0.0000001% population that prefers horrible screeching noises added to any audio content, but arguing for this as an important dimension of subjective variability that we need to incorporate into hardware is I hope clearly absurd).

(2) The vast majority of subjective variability relates to differences in frequency response, which are extremely easy and cheap to adjust in software -- no need to cycle through many multi-thousand dollar headphones or speakers just in hopes of finding one that suits your preferred frequency response target.

(3) Almost every kind of imperfection that can't be corrected via frequency response tuning (e.g. undistorted SPL capability, THD, IMD, off-axis smoothness and similarity to on-axis for speakers, power handling and dynamic compression, etc.) turns out to be almost universally disliked versus when that imperfection does not exist.

(4) Once frequency response balance is factored out either via custom EQ or a "mass market" target curve (which for headphones is indeed more of a compromise, but for speakers is not even remotely a subjective matter, since there are mountains of scientific evidence over decades proving over and over again that the preferred anechoic frequency response for speakers is flat, i.e. the mathematically perfect identity function), which naturally yields a preferred in-room target curve of a smoothly declining slope -- very little (if any) subjective variability remains (e.g. bass boost or treble tilt to taste is one such example).

(5) To be clear, some subjective variability does remain, but it mostly relates to head fit and comfort (for headphones), and for speakers is much more nuanced and tends to relate to beam width and height differences. In both cases, the science is capable of examining these precisely, not in some kind of vague or hand-wavy woo woo sense. Lastly, there is always some need to make some coarse adjustments to bass and treble when listening to older recordings or recordings which were mastered to a non-neutral reference curve (as they should be, else it's impossible to ever reconstruct what was intended) -- but this is not so much a matter of listener preference as it is correcting for mixing/mastering errors in older or lower quality tracks.

For speakers, I’d recommend a pair of Revel M106 (or M105) for ~$1500 new (on sale price, which you can negotiate with dealers usually) plus a ~$500 AVR or integrated amplifier capable of wifi music streaming and HDMI ARC support for TV connectivity, like a Sonos Amp or any of Denon’s AVR products (more on this below).

( ASR measurements + review of the Revel M106: https://www.audiosciencereview.com/forum/index.php?threads/r... )

There are possibly better quality speakers to pick from in this price range (e.g. Genelec 8330A, however this trades off some bass power which you probably don't want unless you have a subwoofer), but to connect these to your TV and streaming music simultaneously requires a more expensive AVR (Audio Video Receiver) that supports unamplified audio outputs, costing around $1K (e.g. Denon AVR-X3700H). It’s ironic because the Genelec 8330A have built-in amplification and digital inputs but don’t accept power-amplified signals, which is significantly more advanced in many ways, yet this advanced nature makes it harder to integrate with TVs and streaming audio simultaneously simply because most AVR products for TV-speaker integration are geared towards “passive” style speakers like the Revel’s for complex and historical reasons. (For some strange reason buying an AVR that allows you to bypass its amplifiers via preamp outputs ends up costing more than AVRs where the amplifiers are always on.)

So for best connectivity in this price range, I’d recommend the Revel M106 along with a Sonos Amp. The Sonos software experience is by far the best at seamlessly and automatically transitioning between wifi streaming music and TV use modes, works natively with Spotify and almost every other music app, as well as having a good app of its own.

However the Sonos Amp (and Sonos Port product which produces unamplified outputs, but for some strange reason omits the HDMI audio input the Amp has, unfortunately, so doesn’t integrate with a TV well) both lack the ability to calibrate bass to your room, which is very important if you plan to upgrade to speakers and/or subwoofers with deep bass, where EQ calibration to the room’s resonant frequencies is essential for best results.

If you want to be a little more future-proof and feature rich, and also be open to multichannel home theater in the future, a modern Denon AVR receiver with Audessy room calibration capability and “HEOS” Music streaming is also really good, but it gets complex as there are a lot of choices and depends on how “future proof” you want to be. And they tend to be fairly bulky.

Note that I’m assuming you want to be able to use the same pair of speakers for both TV output, and for music streaming when the TV is off. This significantly complicates the electronics, and few products get the user experience here right. If you’re willing to manually use a remote or buttons to switch modes, or dedicate the speakers to either exclusively streaming music (controllable from your phone of course), OR TV audio output (which via Spotify TV app can sometimes be sufficient for music too), then much simpler and cheaper options are available.

I didn’t miss it; it’s just not valid. (1) The HD800S is actually not as highly regarded as you may think on ASR because these issues do show up in the measurements and are pointed out in the review (did you read it?), but also (2) the HD800’s issues are indeed very much fixable with EQ, and I can confirm this subjectively with direct experience with my HD800S.

I hope you’re not thinking I’m advocating a crude N-band equalizer to broadly boost the bass and reduce the treble to fix the HD800’s thin-ness and brightness. I’m talking about very precise filters that the community has optimized mathematically to fit the measured response to the ideal Harman target curve: https://www.audiosciencereview.com/forum/index.php?threads/s...

There is no reason scientifically to believe this wouldn’t work, other than the maximum driver excursion and distortion limits of the headphones. Fortunately the HD800S bass is extremely clean and undistorted, and they respond really well here. I’ve tried this with other headphones, and most cheaper headphones do struggle greatly, while the HD800S sub-bass is extremely impressive, clean, and powerful when EQ’ed in this way. And even more importantly, this works wonders to smooth out peaks in the treble that sound too harsh to be prior to EQ.

Even if it couldn’t be fixed with precise EQ though, this is still something we can measure objectively via max SPL and distortion vs frequency tests. For example, another highly regarded headphone is the Focal Clear, and it unfortunately has severe bass SPL / driver excursion limits that rules it out from being corrected in this way.

Of course it’s a fair criticism that such expensive headphones should require EQ at all! And the HD800S are indeed criticized in this way on ASR! Sadly though, it seems virtually all headphones do require EQ to sound remotely near their maximum possible sound quality potential, in terms of their physical capability. I personally love the HD800S for a number of reasons, but one of them is extreme comfort (which definitely is subjective), but I will be the first to admit their sound is painful to my ears until corrected with EQ.

If a $80k watch or turntable makes you happy, then that’s great for you. The reason the science matters here is because most people don’t necessarily have a ton of money to waste, and aren’t necessarily trying to recreate a historical reproduction of some old records (along with all their flaws); most people just want to listen to their favorite music and have it sound the best it can for a reasonable price. This is where the science has the definitive answers, with the potential exception of really old recordings with various quirks due to the lack of well-established audio recording and reproduction science at the time. But digitizing such old works involves some creative art in trying to recreate the authentic old sound when played on modern speakers, and this can be done well (and is done well, in some but not all cases).

But for example, there is no magical difference between a good studio monitor and good audiophile speakers. The differences are well understood and easily measurable. Studio monitors are often more neutral and flat, but also may tend to have a narrower beam, while audiophile speakers often have wider beams and perhaps some coloration like bass boost and treble boost or reduction.

What’s great about the science is it removes all this subjective guesswork. For example, my Genelec 8351B’s out of the box sound too bright for my tastes. But thankfully their built-in DSP allows me not only to calibrate them to perfection to my room, but tune the overall sonic signature to my preference (a gentle downward slope, for a warmer sound). The beauty of good speakers is you can do this in software, rather than buying a different pair of speakers.

About the only thing you can’t tune in software with good speakers is the beam width (assuming there are no flaws like directivity mismatches, distortion, etc. which can never be fixed in DSP). And that’s why e.g. my Genelec still sound different from my Revel speakers: the Genelec are medium width beam, while the Revel are very wide.

Depending on the music genre, this matters. I prefer the wider beam sound on older music and more traditional “audiophile” style music, so for that I tend to prefer the Revels. But for newer music, I and almost everyone else prefers the Genelec’s presentation. It’s important that this is not some subjectivist mystery: it completely makes sense, and through this understanding we are empowered to engineer or select the best speaker for the job without making it unattainably expensive or impossibly convoluted via subjective claims.

For the absolute best, cost is no object? Genelec 8351B + W371A, Revel F328Be, Dutch & Dutch 8C, Kii Three BXT, and many others.

But you really don’t need to pay this much unless you want the absolute best quality AND quantity (capable of going extremely loud, which is an engineering challenge that requires expensive solutions). You may find the “Speaker Attr Comparison” tab here useful to explore the price vs performance spectrum: https://www.audiosciencereview.com/forum/index.php?pages/Spe...

I would recommend learning to read the individual reviewed speaker measurements though, since there definitely is some information lost when compressing these incredibly rich measurements down to a single number, and the science still has some room to improve since there are some dimensions (e.g. the importance of low distortion at high SPL, or beam width) that this fairly old aggregate score model does not currently incorporate.

There definitely is some subjective taste involved, but the important thing is that we can measure these dimensions objectively and understand which is subjective (and why), and which is not a matter of subjective personal taste. We can explore these personal preference tradeoffs without completely surrendering to the subjective nonsense that ‘anything goes’.

For example, it looks like the most subjective factor remaining in speakers is the overall beam width — whether you want a more directional sound, or a more omnidirectional one that fills the room more broadly. This topic is at the edge/fringe of the established science, but the general consensus here is that the objective best would be a multichannel surround sound system of medium beam speakers, where you can either replay recordings meant for multichannel systems, or for traditional stereo recordings at least dynamically select how frontal vs surround you want a particular stereo recoding to be reproduced (just as you can e.g. easily adjust bass boost to taste depending on what you’re listening to, if it wasn’t recorded/mixed the way you like). But given that most people want to minimize the number and size of speakers in their room, there does tend to be a practical challenge of preference here if you can only pick one style (wide vs medium vs narrow beam) for stereo recordings (which while not an ideal format, constitute 99.999% of all music people enjoy).

This trope that ‘the best audio product for you is a highly subjective personal choice’ directly contradicts decades of well-established science, and is a long-standing myth that plagues this industry — and not coincidentally, serves to sell a ton of over-expensive garbage to under-educated (by design, via marketing bunk) consumers.

If you read ASR’s materials on headphone science and the HD800S review in particular, it will become entirely clear why it sounds the way it does, and what you need to do to EQ it to fix its tonal character issues out of the box.

I own the HD800S and if it wasn’t for the scientifically derived EQ filters some community members have developed, I would be entirely unable to listen to them for the reasons exactly as you describe. But once the EQ is applied fixing their FR curve to the Harman target curve, suddenly the sound signature dramatically changes from unlistenably painful (for metal music, for example) to perfectly balanced sound.

I should add though that headphones are definitely more subjective than speakers. This is because we are shoving a speaker 1 inch from our ears and hoping it will sound like the music is in the room around us instead of actually shoved right up against our ear — and the way this sounds actually does differ from person to person due to different ear canal geometry!

This limitation is not the case for speakers, though! The consensus science shows that for speakers, an “objective best” does exist (such that the objective “best” has extremely high likelihood of sounding best to >95% of people), whereas for headphones this number is not as high (though I don’t know that number off the top of my head).

And you know what’s great about ASR? They explain all this! All you need to do is read their resources and understand what the science does and does not help with.

The “Speaker Rankings” tab here is useful: https://www.audiosciencereview.com/forum/index.php?pages/Spe...

Otherwise, let me know your price range and size constraints, and I’ll refer you to the best engineered (with subjective impressions that back it up) speaker you can get. Note that generally to get good bass response you need larger speakers due to the laws of physics, but more exotic construction can push these boundaries, albeit at the expense of greater cost and electrical power demands.

If you want to know what is objectively far better (and subjectively according to most, as the science predicts), look to the Revel F328Be, Genelec 8351B + Genelec W371A, Kii Three + BXT, Dutch and Dutch 8C. These are very expensive, but even in the worst case they are half the price of the B&W Nautilus and perform far better according to what the state-of-the-art science tells us. And subjectively, this is easily confirmed.

I am surprised there aren’t more mentions here of AudioScienceReview.com

For anyone interested in buying speakers (or other audio products), it’s a fantastic resource and an essential one to avoid getting ripped off with severely overpriced and underperforming products; sadly, the high fidelity speaker space is crowded with such products (many of which are borderline scams), using a ton of pseudoscientific marketing babble to push products ranging from “snake-oil” bunk, to mediocre garbage that still costs the price of an exotic car for no good reason.

If you’re curious to cut through the garbage, and learn how to achieve the best sound quality for the best price with a no-nonsense approach, AudioScienceReview is the place to go. They have the highest quality objective measurements on many speakers (that goes far beyond frequency response, before you brush it off thinking that’s what I’m talking about) and tutorials on the well-established science of what makes a speaker sound better than others, and how we design and evaluate this.

It turns out you can get sound quality ~90% as good as it gets for just a few hundred dollars, and ~95% as good as it gets for a few thousand dollars (obviously just rough numbers here). Beware of speakers sold for exorbitant prices and exotic visual designs that tout how they are built, rather than what measured performance they achieve objectively.

B&W speakers are not bad, and I enjoyed mine very much when I had them. But there exist far better speakers at a fraction of the cost, and this includes their high end (like the Nautilus).

Const in C++ is entirely unlike immutability in Rust. Even ignoring the existence of “const_cast”, C++ const is vastly weaker than Rust immutability in several extremely fundamental ways that should be immediately obvious if you’re a C++ professional with even a surface level introductory understanding of Rust. At this point you honestly should just take some time to learn a bit of Rust, and it will likely all make sense. If not, it’s possible you’re relatively amateur C++ developer (no shame in that) lacking experience of the many ways in which C++ const falls very short.

But to directly answer your question, suppose I have a “struct FloatArrayView { float* data; size_t size; };” Of course this is a toy example, but humor me for now and consider the following:

1. Without excessive custom effort (e.g. without manually unwrapping/wrapping the internal pointer), how do I pass it to a function in C++ such that the data pointer within becomes const from the perspective of that function (i.e. you’ll get a compile error if you try to write to it?) Hint: It’s very complex to do this in C++, vs trivially easy in Rust. And no, passing as const or const reference to “FloatArrayView” does NOT work. The only solution in C++ for complex composite types operating with “by-reference semantics” is ugly, complex, and horribly error prone to maintain correctly. For a more concrete example, consider how “unique_ptr” works with the constness of the value it holds. A “const unique_ptr<T>” is NOT a unique pointer to const T. A “unique_ptr<const T>” is, but the relationship and safe conversions between these are not simple or easy to implement or even use in many cases. It gets even worse when you need to implement a “reference semantics” type like this that is inappropriate to be a templated type, or contains a variety of internal references unrelated to the template arguments.

2. Now suppose I solve #1 and pass this into some class method, which then stores a copy of the const reference for later. But I as the caller have no way of knowing this. So after that method returns, I later proceed to modify the data via my non-const reference (which is a completely valid operation), but this violates the previous method’s assumption that the data was immutable (will never change). This creates an incredibly dangerous situation where later reads from that stored const reference to data (that was assumed to be immutable) is actually going to actually yield unpredictably changing results. Const is not immutable. Question: How do you make it so C++ guarantees that passed reference is truly immutable and will never be mutated so long as the reference exists, and enforce this guarantee at compile time? In Rust this is easy (in fact, it’s the default). In C++, it is impossible. The closest you can get in C++ are runtime checks, but that’s nowhere near as good as compile time checks.

Edit: Removed a bunch of perhaps unnecessary extra C++ trivia which I’ll save for later :)

Science must find a balance between extremes in how experimental resources are invested:

Extreme “Exploitation” Bias: “For maximally economical progress, experimental physics must strictly stay within the hard boundaries established by theoretical frameworks.”

Extreme “Exploration” Bias: “For maximally fundamental progress, experimental physics must question all theoretical assumptions and boundaries with extreme suspicion.”

It makes sense to diversify the kind of experimental investments we make along this risk/reward or exploration/exploitation spectrum. There are unique dangers if we weight too heavily on either extreme.

Ironically, this serves as an excellent refutation of the above analogy’s goal (to scoff at EMDrive experiments):

* Imagine telling a mechanic that mounting a ski box on your car with the fat/blunt end pointed forward seems to improve fuel economy vs being mounted the other way.

* The mechanic believes it’s absurd to claim you can improve fuel economy from a change that increases wind drag, so therefore your experiment is wrong since it seems to contradict established theory — case closed.

* Imagine that you more carefully measure, and find that it still seems to work surprisingly well (though also mysteriously reducing braking and cornering performance)!

* The mechanic still denies the result (which is unfair), and questions your test methodology and demands replications of the experiment (which is fair).

* Imagine someone else tries to replicate the experiment with a differently shaped ski box (i.e. not an airfoil, or with the airfoil shape vertically inverted), and fails to reproduce your results.

* The mechanic community concludes that the original result was obviously just experimental error, since the latter failure to replicate aligns with current theoretical consensus of what should happen.

In a similar way, early pioneers experimenting with winged aircraft design were laughed/scoffed at endlessly by established scientists/engineers of the era, ridiculed and crackpots wasting valuable time and resources.

The aerodynamics of the airfoil (which also explains the above ski box phenomenon) is an extremely non-obvious emergent effect, and even to this day is not as fully understood as most people probably assume.

What you describe would be a healthy form of lending (lending out money you otherwise have sitting around, so it can be put to good use). Unfortunately, this is not actually how our global financial systems work.

I think most people would be shocked to learn that when a bank lends out money, they are actually “cloning” their money many times over: each dollar from their reserves actually gets loaned out multiple times over, to multiple people (such that e.g. 100 people all have claim to the same 1 “real” dollar that exists, although it’s not actually tracked that way). But because almost nobody uses physical cash any more in large quantities, no actual cash needs to be printed to cover this. Then, if/when the loan is later repaid (which again is not a physical transaction, but occurs as numbers being updated in balance sheets), the repaid “clones” get “erased” from circulation, and only the real original reserves remain [1]. The banks keep the (greatly multiplied) interest paid though as profit.

You (as a mere peasant) are not allowed to clone your money and lend/invest it out, but banks are (in fact, this is essentially what makes a bank a bank), and it’s called “fractional reserve banking”. Some argue that this is very unfair, and argue that this is actually the root cause of a lot of financial problems and wealth inequality in the world.

For example, this system inflates prices (usually starting with investment assets like land, stocks, etc. but can later trickle down and in very delayed way start to inflate food prices etc. too), due to this expansion of the money supply (commonly called “money printing” even though no actual printing needs to occur) being funneled through financial institutions and into various investment assets and derivatives.

[1] Note that in practice, the total amount of outstanding debt tends to increase over time, because old debt just ends up getting replaced/refinanced with newer debt of even larger size (e.g. due to decreasing interest rates). This creates a system where the global economy is essentially “addicted” to perpetually increasing amounts of debt. The official stance though is that central banks should be able to unwind this debt when we are in good economic health, but this is debated; for example, I don’t know of any case in modern history where overall money supply has actually been decreased without markets immediately panicking and starting to crash (after which the central banks always quickly give in and reverse course, usually re-inflating the money supply far more than even before).

So what is a good “low fructose” (or otherwise healthier) sugar that is recommended over table sugar and HFCS?

I had thought honey was generally regarded as a healthier form of sugar, but now I’m a little confused, since it appears to also be fairly high in fructose (~40%) according to this article:

https://www.medicalnewstoday.com/articles/317728#Benefits-of...

Question: Does this mean honey and high fructose corn syrup are approximately equally [1] unhealthy? Or is honey an anomaly/enigma [2] in this respect?

[1] Apparently honey contains ~40% fructose, which about the same as “HFCS42” (a standard 42% fructose corn syrup product).

[2] I ask this because I’ve read many claims that honey is among the healthiest forms of sugar, with many studies cited that show all sorts of health benefits of honey (at least versus table sugar or HFCS). Is this pseudoscientific bunk that I’ve been mislead to believe?

Here’s the TLDR:

> - The High Energy Electromagnetic Field Generator testing occurred from October 2016 through September 2019;

> - The cost was $508,000 over the course of three years. Around ninety percent of the total - $462,000 - was for salaries, while the rest was used for equipment, test preparation, testing and assessment.

> - When NAWCAD concluded testing in September 2019, the “Pais Effect” could not be proven.

> Despite the fact that NAWCAD says the experiments ended in September 2019 without proving the “Pais Effect,” the inventor asserted to The War Zone in a November 18, 2019 email that his work “culminates in the enablement of the Pais Effect” and that “as far as the doubting SMEs [Subject Matter Experts] are concerned, my work shall be proven correct one fine day…”

However IMO what's really interesting is when we contrast this announcement of test failure with prior released documents showing that the Chief Technology Officer of the Naval Aviation Enterprise wrote a letter to the US Patent Office pressuring them to overturn a prior rejection of these patents, claiming that the technology is operable and that "China is already investing significantly in this area".

Citation: https://www.thedrive.com/the-war-zone/28729/docs-show-navy-g...

I would assume that’s precisely the point of this silly game — to demonstrate directly that you can’t accurately judge someone by their appearance. But maybe I misinterpret?

Yes, I 100% agree with you that “append()” vs “concat()” are certainly no less confusing in their differences than “/=“ vs “+=“ (and the latter are arguably less confusing due to the mnemonic effect you mention).

In fact, to generalize, I think we can fairly say that operator overloads are akin to extremely short function names; in some cases they may work out great when there is not much ambiguity implicit in the underlying problem, but in other cases a longer and more explicitly descriptive name is required to disambiguate. In this case, “append” and “concat” seem quite poor names for different functions, given that they are virtually synonymous and therefore do nothing to describe or distinguish their differences of behavior.

So my claim is that we can do significantly better at resolving ambiguity with carefully named functions (or other approaches) than operators, or tersely/ambiguously named functions (like “append” and “concat”). Of course, this does come at the cost of code verbosity. Just where we should draw the line between too ambiguous vs too verbose, is of course a difficult subjective matter.

I suspect it might have bitten me if I wasn’t so lucky to encounter “/=“ first, which lead me to read the documentation and discover the different kinds of concatenation (“+=“ and “/=“) and their behavior.

To someone just skimming the code though, it might not at all be obvious that “+=“ and “/=“ both exist as concatenation operators, and their behaviors are different.

But yes it’s an aesthetic preference or opinion; I tend to prefer operators only when their only possible behavior is nearly so obvious that no documentation should be necessary. When that’s not the case, I prefer named functions/methods because they permit being explicit about subtle differences.

Allow me to introduce you to C++’s new std::filesystem::path class, which overloads the divide operators "/" and "/=" to concatenate two paths. (Not to be confused with the “+=“ operator, which also concatenates paths, but with subtly different behavior.)

Why did the division symbol get chosen for a concatenation behavior, of all things? Well, I suppose because it looks kind of like how we write paths with slashes, e.g. “dir/subdir/etc”. While I understand how this might be fun aesthetically once you already know the types involved, I find this completely opposite-than-numeric behavior here to be quite ambiguous and unintuitive.

Unfortunately, it becomes virtually universal when a small set of massive companies (with similar censorship ideas) control 99% of all our communications and social media.

I think this evades the spirit of the legal protections here, at least.

Note: I'm not disagreeing with you, and don't really have a solution here. Just pointing out how the current situation feels like dangerous territory.

Wow. I cannot believe I’ve been reading HN for so long and never noticed that. Thank you.

I was not aware that HN had a search feature. Link?

I think the problem here is when buggy “science code” / “prototype code” is used to inform life-and-death decisions made in the real world.

It is therefore rather irresponsible of the scientific community to allow such a bad state of code technical debt to live on for so long, when it’s dealing with a field that is inherently tied to the study of life-or-death scenarios (like we are facing right now).

The only way to avoid this inevitable conclusion of irresponsibility would be if the scientists behind the model warned that it cannot be trusted at all due to the potential of a large number of unknown bugs (did they?). Otherwise, either the scientists are irresponsibility incapable of understanding how to write reliable complex code, or were not honest about the trustworthiness of their code.

Yes I’ve done a ton of research here and have spent a lot of money on speakers (both with good and bad results), so I am happy to answer all your questions.

In my opinion and from much research I believe Ascend speakers (with RAAL tweeters) and Rythmik servo subwoofers are very nearly the best in the world for probably the best price you’ll ever find. Others make comparable and perhaps even better options but they’re generally exponentially more expensive and harder to obtain (not always mass produced).

Even some of the best within 2-3x the price aren’t as good. For example, before I tried Rythmik, the best subwoofer I found was from JL Audio. A JL E112 costs twice as much as a Rythmik F12, yet the Rythmik F12’s sound quality is a league beyond (due to their servo feedback loop tech, the accuracy and precision of their bass is just amazing). And the Rythmik F18 combines massive power (for movies) and incredible precision (for music) in the same product still far below the price of even JL Audio’s entry level stuff.

And other than the Sierra RAAL Towers, my second favorite tower speakers are Revel F206. The Revel F206 are fantastic speakers, but despite costing much more, they still simply aren’t as good as the Sierra RAAL Towers!

I have several speakers from Ascend, Rythmik (subwoofers), Revel, and Neumann. All of these brands are mindblowingly good, but Revel and Neumann are twice the price to get maybe only 90% as good as Ascend. Brands I’ve owned but sold or returned (due to poor performance vs price) include Bowers and Wilkins, KEF, Paradigm, Axiom, and some misc others.

Martin Logan electrostats look very cool and also can sound very impressive but they have two huge limitations:

(1) They’re far from neutral, which means they’ll make some music genres sound great and others worse. With neutral speakers, you can equalize them to suit your preference. But with colored (non neutral) speakers, you often cannot equalize them to fix their flaws (I could explain more why this is, if you want), so neutral is almost always the best choice.

(2) They ONLY sound good if your head is perfectly on axis. If you are seated in precisely the right position and don’t move your head an inch, they can be quite nice, but it’s really annoying to have the sound degrade so horribly everywhere else in the room.

But there is a certain magic to electrostats (when you are in that listening sweet spot) you don’t get from most regular speakers, so it’s a shame they have these flaws. Wouldn’t it be great if a speaker existed that combined the magic of electrostats with none of its flaws?

As it turns out, the RAAL tweeter achieves just that! RAAL’s ribbon tweeter (used in Ascend’s high end speakers) achieves the best of all worlds: these are the only speakers I know that achieves that magical holographic treble precision (detail without being harsh) usually associated with electrostats, while ALSO achieving some of the widest and most consistent dispersion pattern of just about any speaker in the world. The RAAL tweeter is a planar diaphragm very much like an electrostat, so some of the same benefits are achieved but with exceptionally wide dispersion rather than exceptionally narrow.

This amazing dispersion pattern not only makes for the most consistent sound throughout the room I’ve ever heard, but also increases sound quality on axis (because actually the majority of sound heard at any listening position comes from wall reflections, not direct sound).

Note also that all ribbons are not created equal: most of them (e.g. Martin Logan’s lower end hybrid speakers) sound (and measure) worse than a high quality traditional “dome” tweeter. RAAL is fairly uniquely exceptional, and BTW this is reflected in their objective measurements (not just subjective descriptions of “magic”).

Yeah, there’s a good chance your B&W from the 80s sound better than any of Sonos’s speakers, and perhaps even better than modern B&W (since in modern B&W regressed to less neutral speakers to demo a more sparkly sound in showrooms, even though it limits their versatility of optimal music). I’m completely with you about disliking this modern trend of disposable and brick-able “internet required” speakers combined with modern software culture’s bizarre love for this “service” model where you are expected to perpetually accept OTA updates to keep using your product — mandatory OTA updates that are only good until the company behind them stops caring about old products they’re no longer making money off of, and decides to brick or cripple them.

But regarding speakers: There have definitely been advancements in modern speaker design. Probably the biggest advancement is the state-of-the-art science of psycho-acoustics, developed by the Canadian National Research Council, and later continued by the research division of Harman International. The results of this research has directly lead to better speakers by Harman Intl. companies (Revel, JBL, etc.) and throughout much (but not all) of the speaker industry.

Psycho-acoustics informs speaker engineers how to objectively measure flaws in a speaker in ways that go far beyond an old fashioned on-axis frequency response plot (which tells you nearly nothing): measuring FR and others in a densely sampled sphere surface around a speaker in an anechoic chamber has been shown to provide almost all the info you need to evaluate and design a speaker.

The end result of this is pretty amazing speakers for great prices. A $500 pair of JBL 308p Mk II will probably get you 80% of the way to audio perfection for most people, for example. A Neumann KH120A ($1300/pair) will get you maybe 90% to perfection[1], though at the expense of some bass extension which you may want to complement with a subwoofer.

And then we have some exotic speaker designs made affordable; for example, you can get speakers with a RAAL Ribbon Tweeter from Ascend Acoustics which will blow away any other modern speaker costing 2-3x as much. A $2k - $3k speaker and subwoofer system from Ascend+Rythmik[2] will probably get you to 99% of the best sound quality in the world.

For example, I’ve owned Bowers and Wilkins 702 S2 before discovering the science and innovations of some exotic components like the RAAL tweeter. This research later lead me to buy Ascend Sierra RAAL Towers, which while costing significantly less than the B&W 702 S2 sound many times better. (Bowers and Wilkins intentionally choose to ignore the science). The difference (to my ears) is truly astounding — the Ascends are truly in another league of quality and realism. And, the versatility improvement of neutral (flat frequency response, on and off axis) speakers over colored ones is amazing if you like listening to a wide range of music.

[1] https://en-de.neumann.com/kh-120-a-g

The Neumann KH120A is probably the most amazing sounding compact speaker I have ever heard. It is seriously mind blowing how amazing such a small speaker sounds, and how much quality bass it can put out. It also has some of the most impressively perfect scientific measurements I’ve ever seen (which you can review on the link above in the “data and diagrams” section).

The only thing better is Ascend’s speakers with RAAL tweeters, or Genelec pro monitors (which are quite overpriced).

[2] Example: Ascend Sierra Luna pair + Rythmik F12 will probably get you to 98% of audio perfection:

http://www.ascendacoustics.com/pages/products/speakers/luna/... http://www.ascendacoustics.com/pages/products/subs/f12se.htm...

Additionally, notice the strange and perhaps clever wording of “do NOT reflect the company WE ARE AND NEED TO BE”.

It appears they’re trying to have it both ways: They imply that these emails do NOT reflect the company they “are”, but also imply that they DO reflect the ways in which the company needs to be better (which seems a contradiction). So which is it? You can’t be simultaneously innocent and guilty of the same offense.

Imagine if someone committed some crime, and made a statement: “My actions do not reflect the person I am and need to be.” What does that even mean, precisely?

As I understand it, San Francisco reportedly has an even worse problem than Seattle with drug addiction and homelessness, despite more/higher taxes across the board. Therefore, it is not clear that taxes alone will fix this problem; in fact, by purely looking at the data here on the west coast, it would appear higher taxes correlates with worse problems of this sort.

Note that I'm obviously not claiming that higher taxes necessarily cause worse problems, but a negative correlation like this is strong evidence against any claims for a positive causation of (higher taxes -> less homelessness and drug addiction).

There have been many articles and studies written elsewhere on drug addiction and homelessness, so I won't repeat the details -- but my understanding is that the general consensus is that solving this requires policy change more so than funding change (but obviously both are necessary for the best results).

This is interesting because the idea of owning a car that takes regular OTA updates, and can be disabled or tracked remotely at any time by the manufacturer — to me is very scary/creepy.

Maybe I’m paranoid, but I have a hard time not extrapolating the OTA update experience from other software (desktop, phone OS’s) to cars too: updates breaking features inadvertently but with no prompt fix, injecting new spyware against your will but with no real alternate, updated terms of service that you have no choice but to agree to if you want to be able to drive home, etc.

Given the track record with OTA software updates everywhere else, we need positive assurance that the there won’t be these same profitable compromises of user experience, privacy, freedom, etc. eventually for these cars. Because the status quo is quite the opposite. Have you read the terms of service and related documents you signed when buying your Tesla, in full?

I fully agree about self driving / driver assistance though, and for example am thrilled that Toyota now includes in all their new cars (for free, as a base feature) their entire autonomous drivers assistance sensor suite and features: adaptive cruise control (works in stop and go traffic), lane steering assistance, automatic braking, automatic high/low beams, etc. (and it all works very reliably in my experience). I know it’s nothing like the autonomous features Tesla offers in their upgrade package, but this is a very welcome change in the industry.