I’m curious if anyone knows how Google internally works around the lack of support for gRPC in the browser? Perhaps gRPC is not used for public APIs?
The lack of browser support in the protobuf and gRPC ecosystem was quite surprising and one of the biggest drawbacks noted by my team while evaluating various solutions.
In the spirit of curiosity and playfulness I would like to respond with a bold claim:
The distinction between self and other is simply a concept. Certainly useful in many situations, but ultimately just one perspective - one tool at the mind’s disposal.
One could reasonably argue this is the central teaching of Buddhism. Alan Watts helped articulate these ideas to western audiences in many of his lectures:
“You see, the point is that an enormous number of things are going on inside us of which we are not conscious. We make a very, very arbitrary distinction between what we do voluntarily and what we do involuntarily, and we define all those things which we do involuntarily as things that happen to us rather than things that we do. In other words, we don’t assume any responsibility for the fact that our heart beats, or that our bones have such and such a shape.
If you become aware of the fact that you are all of your own body, and that the beating of your heart is not just something that happens to you, but something you’re doing, then you become aware, also—in the same moment and at the same time—that you’re not only beating your heart, but that you are shining the sun.”
I’ll bite. The data/graphs look suspicious and convincing enough, so WTF did happen?
I was hoping for an answer but alas there was none. Anyone have plausible theories? Is this an unexplained mystery or just an artifact of S curve growth?
> to produce one litre of petrol it takes 1.29 kg of oil, of which 85% (1.1 kg) is carbon. And as only 1/10,750 of the carbon remains from the plants that were buried millions of years ago, our one litre of petrol is the result of 1.1 x 10,750 = 11,825 kg of carbon from ancient plants. Finally, as plants are approximately half carbon, that means that 23.65 tonnes of plants were required to make just one litre of the petrol available at your local station
Thanks! I was going to make a similar comment myself but HN isn’t too keen on these ideas yet. Glad to see other like-minded folks here.
The Wikipedia page for plasma cosmology is extremely biased, so I suggest anyone interested in these topics should read the previous comment I made and some of the resources I linked there.
This is a thoughtful comment but IMO your phrasing is too strong and potentially misleading.
As far as science is concerned, the jury is still out on stellar nucleosynthesis. The leading theory - that gravitational collapse causes fusion near the core - may be entirely incorrect, and there is mounting evidence suggesting so including a plethora of modern observations that the theory fails to explain.
For example the so-called ‘coronal heating’ problem has existed for 75 years, since the solar corona was first demonstrated to contain plasma with temperatures of 1 million degrees kelvin and above, much higher than the photospheric surface temperature of approximately 6000 K [1].
As another stunning counterexample, current gravitational models do not explain observed properties of the solar wind like spatial variation and periodicity in time (thanks voyager 1 & 2), among many other things.
See [2] for a dense intro to more unresolved problems in solar physics.
Alternative (non mainstream) theories like “plasma cosmology” are gaining traction in the scientific community for this reason. Note though that many papers do not use that particular phrase to describe their work.
Unfortunately (for intellectually curious folks), the contributors to Wikipedia pages on these physics topics are extremely … biased. Reading [3] may leave you with the impression that “plasma cosmology” has been completely debunked by the scientific community, when in fact nothing could be further from the truth ([4][5] are examples to support this claim).
JS engines have had many serious vulnerabilities in their JIT optimizers, it’s not myopic at all and is a well known technique in the industry.
I agree that some folks aren’t executing untrusted ruby code so they wouldn’t have to worry about this - but how many PaaS/SaaS products out there are? Or how about third party dev tools that are blindly downloaded and executed on local workstations or CI pipelines?
Adding to this great comment with my own experience at work where we extensively use signal processing and time series compression algorithms to visualize large amounts of biological signals in the browser.
WebAssembly is less than 10%-20% slower than native code in our benchmarks and tests for algorithms like fast wavelet transform and bit packing.
WA allows us to aggressively optimize ahead of time when compiling, and be less sensitive to JS engine performance pitfalls around JIT optimization and garbage collection.
> we develop the notion of representation group flow (RG flow) to characterize the propagation of signals through the network. By tuning networks to criticality, we give a practical solution to the exploding and vanishing gradient problem
I am surprised and a bit disappointed this paper does not mention mean field theory or dynamical isometry at all.
Mean field theory applies methods from physics - namely random matrix and free probability theory - to derive an exact analytical solution for information flow through a neural network.
It turns out that simply initializing the weights of a plain CNN using a delta-orthogonal kernel allows all frequency components (Fourier modes) to propagate through the network with minimal attenuation. Specifically, networks train well when their input-output Jacobians exhibit dynamical isometry, namely the property that the entire distribution of singular values is close to 1. This technique effectively solves the exploding/vanishing gradient problem.
The impact is shocking: the time to train a NN to a given accuracy becomes independent of network depth. No tricks like batch normalization, dropout, or anything else are needed. This insight has been proven for a wide range of architectures from plain FFNs to CNNs, RNNs, and even transformers.
I highly recommended reading the papers “How to Train a 10,000 Layer Neural Network” [1], and “ReZero is All You Need: Fast Convergence at Large Depth” [2].
It’s a valid concern and many folks have raised it here on threads concerning this topic.
The short answer is yes: beyond mannitol, niacin, and even bespoke substances designed to mimic the organoleptic properties of ayahuasca have been used effectively as placebos.
At modest dosages (above the threshold of perception but within the therapeutic window) the side effects can be surprisingly similar in the general population, and thus difficult to distinguish for the average joe.
Of course studies vary widely in quality and rigor, so one must critique experimental design on a case by case basis.
Listen, your comments on all these threads make it clear you’re seriously pessimistic about psychedelic medicine, and that’s perfectly ok.
What bothers me is the incessant pushing of this agenda:
> anyone ending up in a psychedelic research study is doing so because they think psychedelics are promising for mental health and they want to try it out
> maybe the real reason psychedelics show efficacy in depression is that they are a sort of super-placebo
This amounts to uninformed armchair dismissal that is inconsistent with a growing body of scientific evidence, and easily refuted by a cursory survey of recent literature and clinical trial results.
Surely you’re familiar with double blind randomized placebo controlled studies? I’ve linked you to recent ones in other comments.
These methods aren’t perfect and your concerns certainly have merit in general - I just don’t understand why you continue to make these broad and suggestive claims without any supporting evidence.
I'm afraid you're misrepresenting this important area of research.
> Psilocybin is being studied in conjunction with 10-20 therapy sessions
For a quick reality check, you could take a look at Compass Pathways and their phase 1 & 2 clinical trial results [1][2]:
> we have completed a phase IIb clinical trial of psilocybin therapy for TRD, in 22 sites across Europe and North America. This was the largest randomised, controlled, double-blind psilocybin therapy clinical trial ever conducted, and our topline data showed a statistically significant (p<0.001) and clinically relevant improvement for patients who received a single high dose of COMP360 psilocybin with psychological support
That is a single dose along with a single therapy session, showing positive benefits lasting 12+ weeks. For treatment resistant depression (TRD), a notoriously difficult condition to cure and one that affects many people.
There are literally no other drugs on the market that come close to the safety, efficacy, and durability profile that is being observed!
Do you think the FDA granted breakthrough therapy status in 2018 because of some collective "utopian fantasy"?
It's well known there are numerous indications for psychedelic assisted therapy showing great promise - from anxiety and addiction to end of life care [3]. It's not just psilocybin either, many of these psychedelic substances are pharmacologically related - so they act on similar receptors and areas of the brain, and thus have similar therapeutic profiles [4].
Let's be realistic. Your appeal to extremes is an uncharitable interpretation of my comment, and frankly your uninformed pessimism on this topic is a disservice to anyone interested in learning about or benefiting from these advances.
I know you comment on a wide range of topics here, and perhaps you’re jaded by some personal experiences with local ketamine clinics - but IMO your tone comes across as unnecessarily pessimistic and a smidge anecdotal.
Ketamine/MDMA/LSD/Psilocybin and their derivatives are on their way to being recognized as the safest, most effective, and widely available medicines - for nearly every common mental health disorder - currently known to humanity.
Yes there will always be predatory behavior in healthcare, and regulation can help. And of course no medicine is 100% side-effect free or effective for everyone.
But shouldn’t we be complaining that many of these substances continue to remain federally outlawed in the US despite a huge body of scientific evidence and multiple companies discussing phase 3 clinical trials with the FDA this year?
I don’t really hear many folks calling for tighter regulation on these substances, especially given the current regulatory situation.
While the benefits of ivermectin in a single-drug regime are heavily contested, a lot of folks here seem to be unaware that certain multi-drug regimes have shown promising results. For example see the positive outcomes of this double blind placebo controlled RCT evaluating ivermectin with doxycycline.
> Our trees accumulate metals from the soil, making their wood less digestible to fungi
Given that fungi play a crucial role in decomposition and also water/nutrient exchange between living plants, do you think there are any major risks from second order effects? Particularly if these plants become well established and dominate a local ecosystem?
And as a follow up question, does the increase in metallic content pose any extra risks when burning in a forest fire? Perhaps through emissions or thermodynamic effects.
Very interesting idea, good luck and thank you for the thoughtful commentary here.
Connect-Web: TypeScript library for calling RPC servers from web browsers
https://news.ycombinator.com/item?id=32345670
I’m curious if anyone knows how Google internally works around the lack of support for gRPC in the browser? Perhaps gRPC is not used for public APIs?
The lack of browser support in the protobuf and gRPC ecosystem was quite surprising and one of the biggest drawbacks noted by my team while evaluating various solutions.