tbenst

Your request is admirable! Here’s how I go about gathering info:

Google “covid 19 origin evidence”, look for academic publications or scientific journalism that is well-cited & from reputable sources, eg

[1] https://www.nature.com/articles/s41591-020-01205-5 [2] https://www.latimes.com/science/story/2020-05-09/was-the-cor... [3] https://www.cidrap.umn.edu/news-perspective/2020/05/scientis... [4] https://www.nationalgeographic.com/science/article/coronavir...

We really need to do better with scientific communication. As scientists we are evaluated too much on our communication with other scientists (ie paper publishing), while communication with the public is not weighed much for career advancement. I wish this structural problem would be discussed more so it can be addressed.

But not all of this is on the scientists. The public must do better. We can’t just blindly trust what a senator says on Fox News for political expedience, or “trust our gut”.

No, it is not a coincidence.

There are a lot of bats in Wuhan. There are a lot of bats carrying coronaviruses. Coronaviruses have triggered past epidemics. Ergo, there’s an institute for virology in Wuhan.

Listen starting at 6:30 in the podcast I posted from Nature. There is indeed strong correlation but no causal relationship established.

The author is irresponsibility propagating a conspiracy theory and elevating its status in the public’s mind.

I’m a bioscientist. It’s frustrating to respond with evidence and in good faith, and be downvoted by those who simply disagree. But sadly it appears that the loudest voice prevails over reason.

https://www.nature.com/articles/d41586-021-00599-7

On the contrary. A leak implies that something was contained. Notwithstanding the complete lack of evidence for a leak—and one could waste a lifetime trying to disprove claims that have no evidence-if of natural origin, the virus was already infecting animals and/or people.

This article is written by a journalist who is clearly knowledgeable about safety practices and mistakes in US labs, but does not consider the extensive knowledge we have about the sequence of SARS-COV2. The preponderance of evidence supports a natural origin of the virus.

This is no way exonerates the Wuhan government from possible culpability—indeed government officials did deliberately suppress information—but this investigative opinion doesn’t pass scientific muster. Misinformation.

https://www.nature.com/articles/s41591-020-0820-9

The players weren’t allowed to sweep.

Ha, funny to see the resolution to my bug report (among others) on front page of hacker news, very grateful of SRCC’s work to resolve!

I had this initial impression with PyCall from Julia, and would point to my own Conda env. But I quickly discovered when I ran my code on another machine, that it was a bit annoying to have thought I installed all the dependencies only to find the python ones missing. The auto install addresses this use case (for distribution/ sharing).

GPT-3 can’t play Go.

Thank you for the history lesson! I had no idea IPv4 was so old

This is almost certainly false. For example, see the arrest of the leader of the Proud Boys (for events prior to Capitol, but played active role on Parler encouraging proud boys violence that day):

https://www.pbs.org/wgbh/frontline/article/several-well-know...

False equivocation. This did happen but Twitter promptly banned this and variants [1]. Parler is failing / refusing to moderate.

[1] https://www.theguardian.com/us-news/2021/jan/10/hang-mike-pe...

A bit of an odd letter, since Warner’s letter isn’t asking for documents but rather establishing a “duty to preserve” [1] due to high likelihood of impending litigation. IIUC, the senator’s letter is about all that can be done while the Senate is not in session.

Insurrections are not protected free speech, and Gab may have liability. No doubt this will be litigated, and if Gab were to delete content now (as opposed to removing access), this might constitute obstruction of justice.

[1] https://www.kramerlevin.com/images/content/2/5/v4/2562/Reaso...

Before installing NixOS, I’d recommend anyone on Linux to give home-manager a go first for the userspace declarative approach with Nix. After a year with full-time NixOS, I switched back to Ubuntu + home-manager as best of both worlds—declarative when feasible, imperative when convenient.

[1] https://github.com/nix-community/home-manager

Anandtech / Tom’s Hardware, if you’re there, please do this!!

Totally fair, edited

SIMD and ie Nvidia WARP are not the same. Idk about Apple’s GPU, but for example there is no GPU alternative to the SQRTPD instruction (Square root of double precision). Also, when there is branch divergence across threads, CPUs still do a much better job than GPUs.

Curious to think about how unified memory may change the ratio of flops/memory access when it makes sense to shift job from CPU (better for low number) to GPU (better for high ratio)

I think we effectively agree with each other and the author. The M1 is better day-to-day processor but not a AMD/Intel killer for (edit: some) compute-heavy workflows...(yet?). Discussion more pertinent for Mac Pro, where current M1 would be worse than last gen Intel for some common workflows on those devices.

For many linear algebra heavy workflows, (numpy, R, Julia, etc.) I expect that AMD and especially Intel processors with AVX-512 will crush the M1 on real-world benchmarks. But this isn’t a reflection of RISC vs CISC, and Apple could choose to add hardware acceleration for wider instructions and hopefully will in the future.

Slack fosters vibrant (free) communities in open source and academia. I fear these are not long for this world.

Ha! The technology itself is cheap, fabricating a bunch of de novo mRNA is only a few hundred dollars these days. I’m not sure how much volume is needed for treatment, and there’s obviously much stricter QA/QC that must happen for human use, but still the actual manufacturing part is likely no more than single digit thousands per dose. At scale, per your semiconductor analogy, could likely be brought down to single-digit dollars with time.

Thanks for creating and sharing this! Always easier to criticize than create. I for one think your code reads as quite pythonic. Using convenient python wrappers for c libraries is the raison d'être for python in the first place. See numpy, pytorch, ...

This is a kind and thoughtful reply.

I would push back on the notion that there has to be a cause for depression. Maybe there is, but it can lead to the dangerous conclusion that it’s somehow one’s fault, when it’s actually a maladaptive behavioral state; a disease.

Fully agree with your advice to talk to a professional.

I was specifically referring to the Pope story in that case. I have no idea if the Hunter Biden laptop is actually real, but our default assumption should be skepticism: https://www.washingtonpost.com/outlook/2020/10/24/hunter-bid...

Unfortunately, tinnitus is usually caused by hearing loss which comes from the death of neurons in the ear, and we are a long way off from finding a solution. But to the extent that the percept of ringing comes from recurrently firing ensembles in the brain, than neuromodulation techniques may bring relief from the perceptual aspects.

TMS seems modestly promising for tinnitus, but experiments thus far have been poorly controlled so hard to tell: https://www.nature.com/articles/s41598-019-48750-9

Our lab has done work with using optogenetics for inhibiting neural activity, and I wonder if the inhibition approach might be better than disruption. It’s a bit hard to figure how TMS does inhibition vs excitation, even with electrical stimulation it’s complicated and varies by celltype!

I passed along this thread to Ritchie, so hopefully he can chime in as he has time. We actually talk quite a bit about Neuralink, and have colleagues that work there.

Speaking for myself, I love to see the resources increased in this area of research, but am nervous that expectations are too high. Brain surgery for a consumer product seems pretty far off and the early applications will almost have to be medically motivated. That said, Neuralink seems to be doing a phenomenal job in productizing and further developing off of current electrode techniques. I hope the technology becomes available for the research community!

Re noninvasive human neuroimaging, fMRI, fNIRS, MEG, or EEG are your best bet today. fMRI is probably the best of these in terms of what can be decoded from thoughts/perception, and has the highest spatial resolution. I think these have come down in size as to fit in a van, but certainly not very mobile and require powerful magnets. They also capture a proxy of neural activity, the blood oxygenation level, a proxy for neural activity.

ipad dying so will try to update this comment in a bit...

Edit (continued):

> Could ML / signal processing increase the resolution of EEGs? Could we ever extract video signal from an EEG?

The skull acts as a low pass filter and most of the high information content resides at higher frequencies. This seems to make it impossible to decode, say, speech or imagery from the brain’s activity. No ML can save you from a low pass filter.

> Can an EEG induce perception into a brain?

EEG is a recording technique, but tCDS has some mixed evidence for increasing alertness. Transcranial magnetic stimulation definitely can trigger a percept if done over visual cortex, but it’s not very fine in detail. Targeted ultrasound may do better in the future.

> With regard to optogenetics, what are the things currently being studied with the technique? What pressing questions can it be used to answer? Are there any improvements to the technique that could push it further?

So much to say here, will direct you to our lab website for a brief sampling https://web.stanford.edu/group/dlab/publications.html

I’m not too familiar with infrared neuromodulation, but it isn’t a widely used technique in systems neuroscience. My guess is it would be hard to target individual neurons, let alone target a genetically defined population of cells. Heating is usually what causes damage from light stimulation, so we try to minimize heating as much as possible—the brain very tightly controls temperature

For the mouse or rat brain, most everything :). For primates/humans, the skull is a lot thicker, but perhaps cortex is still reachable / longer wavelengths of light might be able to penetrate deeper (multiphoton techniques)

Karl has said publicly a few times that he sees optogenetics as a tool for basic scientific discovery, that then inform treatment of human patients, ie where to use transcranial magnetic stimulation. Any method that requires genetic manipulation in humans will be tricky. PNS seems likely to be the first application in humans, although there are also clinical trials going in the retina, not sure how they’ve progressed though.

You need an actuator to induce current into the neuron. In this case, a light-sensitive ion channel, which have proved highly effective.