"A few days ago, a peer-reviewed research article was published online by the science journal Cell, one of the world’s top molecular biology journals. The article was authored by pro-vaccine researchers at Stanford University and elsewhere. As part of their research, the researchers tracked how long mRNA from the vaccines persisted in the body.
Contrary to Collins’ previous assertion, the mRNA did not disappear in “a few hours,” a few days, or even a few weeks. In fact, mRNA from the vaccine persisted in a person’s lymph system some two months after vaccination. We actually don’t know how much longer it lasted because the researchers only tracked the mRNA for that long."
At 13:24: "A grand oracle system so that we can implement oracles with a reasonable guarantee it won't just ultimately come down to opinionated validators or random oligarchs that have been elected to the council"
Instead...:
"A jury elected on objective real-world criteria... they'd operate as a offchain body"
"... some game theory as to why they would actually tell the truth ... a combination of professional liability w/ Michelin points and economic incentives"
Moved to Go after being frustrated with Python and I couldn't be happier. I know there are complaints with the language but I'm not that picky, I can get in and hack features together really quickly. Deployment is awesome and easy. Edit: I'm specifically referring to cloud development.
Not the poster you're asking, but in my case, have used Python extensively for cloud-related programming, lots of AWS lambda. Currently in a special kind of hell trying to move our company's lambdas from Python 2 to Python 3. It should be easy but there's an internal Python 2 library that is in everything that is massive and almost impossible to excise without significant effort. Started writing lambdas in Go, you ship your compiled binary and the lambda environment will never care what version you're running. Shipping a binary means I'm no longer worried about my build environment and pip and shipping all my Python junk the right way. Also in other areas, simply transposing some code to Go has had huge performance benefits over Python and concurrency is just so easy to pull off.
> Shipping a binary means I'm no longer worried about my build environment and pip and shipping all my Python junk the right way.
I'm not sure it is that clear a win: for go, if you want to be able to reproduce a binary in future, you would still need to keep track of exact versions of all your build-time dependencies such as the compiler toolchain & all library dependencies, and version control or archive backup copies of these. You still may need to figure out exactly what version of each component you're running if serious security vulnerabilities are found in some component.
I agree it is much simpler and easier to ship a single statically linked self-contained go binary than a python app, but both languages arguably require a similar amount of effort to having a reproducible process to build and deploy.
Shipping is simpler. I'm using the Go module system right now and I haven't had any problems so far. I keep library usage to the minimum, mostly I've gotten by with only the standard library.
We already knew it way before the study: mortality rate Covid-19 = 1.5%. 40.000 deaths. 100*40000/1.5 = 2.600.000 infected => 5% prevalence (Spain population 47 million)
Based on NYC mortality figures as a lower bound, I also thought the rate must be in that neighborhood.
But there were influential dissenters, e.g. the infamous Stanford study a few weeks ago estimating nearly 3% infected in Santa Clara County, or the Swedish authorities’ conviction that 25-30% of the population in Stockholm is infected.
This study seems to provide pretty strong evidence that those estimates were too high... not that I expect it to change anyone’s mind at this point.
The mere fact of getting an alert of a possible positive contact would lead to stigmatization. Under a de-escalation scenario with social distancing measures, it is not difficult to find out who was the source of contagion, no matter the privacy protection scheme of the app. This is a faulty design of any contact tracing app.
The goal of a contact tracing app is to incentive self-quarantine of new contagions. But the possible outcome taking into account social effects is stigmatization. The error of this approach could be letting positives being out of confinement spreading alert events and virus at will.
An immunology passport app is the resource needed to avoid this situation, guarantying social environments are free of potential sources of contagion.
Massive testing and immunity passport is the tool to find out and isolate infected people while avoiding stigmatization and privacy violations. Turning neighbors into prosecutors is not a good idea.
