andrewon

- Don't depend on other people's software services. - Buy a domain and host your own website.

I had the exact chain of thought, only to find that traffic of the site I built is at the mercy of how Google decided to rank webpages, and putting AI > youtube > Reddit in front of everything else.

- Mailing lists beat social media accounts.

Similarly, Google set the metric for what counts as spams. Your emails can all go to the spam folder if their AI decides it should.

A primer to Depth model in Stable Diffusion 2: https://stable-diffusion-art.com/depth-to-image/

On the side.

Thanks. I now have a few apps in the app store. I want to try spending resources growing them.

This is surprisingly useful. Thanks!

I used an iPhone app with similar interface but couldn't find one that's as intuitive on web until this one.

Can this be run without colab Pro? (High vram)

After lastpass, I am a little wary about password managers. Not sure when they plan to "monetize" me.

Been using google password since lastpass turning on paywall. What's your opinion of Bitwarden? Reasons to give it a try? Having all passwords locked up is no fun.

They should do the same for patent. I see companies and now universities freely take tax payer money to develop their own products. It just becomes an additional source of funding with little string attached.

It's fine to use tax dollar if there's potential for public good, but the tech developed should be released to the public domain right away. I have seen a selfless act from an academic group making decision not to patent a technology because they felt its an important one that many other things can be built on. Now tens of companies were started based on that tech and counting. I hope our gov understand how much value can be unlocked by public domain technology.

The shortest time from arrival of the blood sample in the laboratory to the initial diagnosis was 7 hours 18 minutes.

After establishing a diagnosis in Patient 1, we updated our bioinformatics framework to permit the transfer of terabytes of raw signal data to Cloud storage in real time and distributed the data across multiple Cloud computing machines to achieve near real-time base calling and align- ment, a step that reduced the postsequencing run time (base calling through alignment) by 93%, from 7 hours 21 minutes to 34 minutes.

Amazing basecalling and alignment of a human genome can be done in 34 minutes. How far we have come since the first reading of human genome.

What's wrong with letting employees to hear both sides of the argument, from professionals? Both unions and "consultants" are resourceful organizations. Perhaps they can keep the good and take out the bad practices in their unionization, if they ever choose to.

Yes R&D is cost. Sales is profit.

You are right. The base salary could be similar to tech but bonus and stock are generally way lower. My company has problem keeping SWEs.

I am in medical industry. Once I couldn't fathom why a company making a messaging app could think they were "changing the world". It dawned on me that value and meaning can be relative. You can always spin the arguments and convince yourself what you are doing is meaningful, or as least to your employees. For example, didn't Amazon empowered consumers very wide choices that's previously not available? Didn't they created tens of thousands jobs across diverse geographical areas?

The progress in artificial intelligence in all categories is fascinating and is in the modern day what electricity was at the turn of the century

I personally don't buy into this analogy which I think its just a marketing ploy. The advance in AI is incremental in reality. Before neural networks have been widely implemented, other machine learning algorithms had been working fairly well. The highly non-linear models with hand-crafted domain-specific constraints just fit the data better and push the accuracy higher. But fundamentally, the way we treat data and prediction is not very different from other machine learning models.

This sounds like a good idea to reduce monopoly. Any downsides? Would love to hear arguments against this.

They do work on an extraction flowcell that can be added on top. I’m hoping they can make it as easy as adding sample to a well, at least for blood or saliva.

It's a fascinating interview. Haven't heard about him before but his academic background looks legit, albeit more on cardiology not infectious disease.

https://scholar.google.com/citations?user=LzqEaOkAAAAJ&hl=en

He made many extraordinary claims in this video. Here are some searches I have done on his claims.

# Claim: You cannot be reinfected by covid-19

This Qatar study confirmed reinfection:

[Efficacy of Natural Immunity against SARS-CoV-2 Reinfection with the Beta Variant | NEJM](https://www.nejm.org/doi/full/10.1056/NEJMc2110300)

They even studied the reinfection rate difference between alpha and delta variants.

# Claim: There's no FDA approved vaccine

Pfizer vaccine is fully approved for use in 16-year and older on Aug 23, 2021. Below is the press release from FDA

https://www.fda.gov/news-events/press-announcements/fda-appr...

# Claim: Effectiveness of Hydroxychloroquine is dismissed

Hydroxychloroquine is actually the most studied drug against covid-19.

https://www.statnews.com/2020/07/06/data-show-panic-and-diso...

We can always argue that the dosage, setting, timing, etc are not right in these studies. To me the fact that we don't find it particularly effective in this many studies is a good reason to give other drugs a try.

# Claim: Medical community does not focus enough on treatment. If properly treated, majority of deaths can be prevented.

This is a fascinating point. I don't know how patients are being treated. But there must be some improvement in treating them as the death rate is way down even in unvacinated.

He criticized hospitals applying arbitrary criteria for monoclonal antibody treatment. I see there's a shortage in the drug so it makes sense to only use on high-risk patients, e.g. old, unvaccinated.

But I do like his point that we should focus more on treating patients.

# Claim: Myocarditis induced by vaccines is a lot more common and severe than claimed

This is a more troubling one if true.

There's quite a few studies on the rate in multiple countries. All appear to be in line with the CDC number, e.g.

https://www.nejm.org/doi/full/10.1056/NEJMoa2110737

The million dollar question is what's the rate of severe Myocarditis vs severe covid in children and young people? This should be the basis of cost-benefit analysis of mass vaccination.

# Final thoughts

His prior academic credential appears legit. He talks in a way that's fact based. He must have taken enormous personal risk in speaking out against the establishment. I want to believe him but so far many of his claims do not check out based on my own research. He quoted many studies in the interview. Would be helpful if there's a reading list to better understand his argument. The interview is more like his monolog as Joe has no expertise to critique him. It would be interesting to have someone equally knowledgable but hold an opposite view so we can hear the arguments from both side.

Research article https://www.nature.com/articles/s41592-021-01317-x#Sec28

That reminds me of the idea that brain as a predictive engine that constantly refine its internal model to match anticipated sensory outcome. In this case the inference is way more complex based on cognition that the placebo would help relieve pain.

Based on this idea, however, the study does not identify the key pathway responsible for placebo effect, but only the result of it. If the brain anticipate relief of pain, the parts related to pain would activate. If it anticipates other effects, other parts of the brain would do.

Also this one... "Racial profiling is at the heart of Amazon Ring’s business model."

There's a liberal tendency to hyper-focus on race and see everything in the lens of race only. Ironically, I don't think this helps racial injustice that they are set out to resolve.

"Training was done on eight NVIDIA A100 GPUs for 1000 epochs" "trained on four NVIDIA V100 GPUs for 300 epochs" "model was trained on a NVIDIA DGX-1 cluster using eight GPUs"

No doubt these are model innovations. But can't help wondering how much ready access to arrays of GPU might give them advantage... You know how many different models need to be tested before getting to the winning ones.

The drop after switch makes sense. There's gotta to be significant number of people won't switch to Spotify just because of Joe Rogan's show. Half of them is about right.

His show is just one of many in my queue. I won't go through the troubleshoot of using two apps for podcast just because of him.

" the downside is that we pay more for things than people overseas do"

Isn't that they just force you to buy extended warranty on everything?

Would have to disagree. They will need to confirm these were not artifacts. Long read sequencing assembly is the best approach to show the computationally assembled reads indeed coming from single contiguous molecules.

Was thinking the same and wondering how did they obtain the million-base long sequences. Turned out they used short read sequencing with 150 or 250 bp reads and computationally assembled the long reads [1]. While this is a traditionally valid method, the newer long read sequencing technologies such as Oxford Nanopore or PacBio would be an more appropriate and direct method.

[1] https://www.biorxiv.org/content/10.1101/2021.07.10.451761v1....

While Uber and Lyft are selling their autonomous driving units, Baidu launches driverless service. Have they figured out how to do it right or just regulatory?

Several considerations here:

1. If holding off AstraZeneca vaccine does not affect vaccination rate during the investigation period, it is a prudent thing to do.

2. If vaccine rate is expected to drop, number of expected increase in death per day due to covid vs blood clog should be compared, if we were to minimize short term death.

3. More concerning is unknown effect that could take a long time to materialize. This is a tough call to make since any effect is only theoretical at this point.

The downfall of building an entire testing system based on a proprietary technology. Other things can also go wrong, like manufacturer refusing to develop a test on it because it is not profitable, priority in developing a test could cause delay.

The setup works both way.

They already had shares that scheduled to sell but decided not to execute. It seems like the advice I always got from my company's legal department when I asked if it was ok to something: don't do it.