rocha

But isn't this inefficient since the agent has to "bootstrap" its knowledge of the new language every time it's context window is reset?

The number of people at that compensation level is very very small, and they will probably go for a O1 type visa instead of a H1B.

Yes, but you will get an error since string and templates are different types and have different interfaces.

Great resource! Thanks for bringing it up.

For some of the card games I am familiar with (Netrunner, MTG) there are resources online (wikis, dbs, deck builders) that have all the graphic assets. Alternatively, you can scan or photograph each card in your deck, but that may be time consuming.

As for software, there are engines out there that do content-based image retrieval (CBIR) out of the box [1]. It should be possible to build something quickly in OpenCV. You may be able to get away using simple image template matching by putting a few constraints on how the camera sees each card. Something more robust can be also be build using image descriptors and approximate nearest neighbors, as in [2].

[1] https://en.wikipedia.org/wiki/List_of_CBIR_engines

[2] https://blog.francium.tech/feature-detection-and-matching-wi...

I think it is very doable for card games with thousands of different cards like MTG. I remember there was an old school SIFT/SURF image retrieval system that was able to deal with tens of thousands of album covers super fast. That was more than 10 years ago IIRC.

Another option is an image recognition system trained on already existing cards. No need for special ink or qr code, just a camera as the reader.

The walk really slowly. The can "run", but only for a very limited amount of time before their servos overheat and the dog has to take a nap.

Here are the arguments that the WHO used in 2015 to change the practices for naming new infectious diseases:

https://www.who.int/news/item/08-05-2015-who-issues-best-pra...

This is a nice thread on the subject: https://twitter.com/nataliexdean/status/1310613702476017666?...

Nassim Taleb wrote a whole book (Skin in the Game) under a similar premise: “Never trust anyone who doesn’t have skin in the game. Without it, fools and crooks will benefit, and their mistakes will never come back to haunt them.”

I find https://adguard.com/ very handy for those cases.

One thousand in a million is the same as one in a thousand.

I agree completely. It is very easy to dismiss the problem if you are young and you only look at your individual risk. But if you look at society as a whole, the problem is significantly more complicated.

You said there was no data available, and I provided a reference.

The data you cite has a large age group as well (0-69), which has the same problem you describe. See the comment by user kmm below for a reference with a better breakdown of estimated IFR by age groups. The reference also shows how the IFR increases exponentially by age.

If you want to see what number applies to you in particular, then you need an specific breakdown. But if you need to see what is the risk for the population in general, then the estimated total IFR, sampled from that same population, is valuable. Think of individual risk vs systemic risk.

You are correct. Thanks for pointing it out.

Your reference is very good. It is still a pre-print, but the breakdown is very informative. It estimates the IFR (Table 4) for ages between 0-34 as 0.01% (1 in 10,000), increasing exponentially from there. The estimated IFR for the next age group, people between 34-54, is between 0.04% and 0.2% (1 in 2,500 to 1 in 500), one to two orders of magnitude greater. For people over 85, the IFR is 36.8% (~1 in 3).

Note that the population over 34 is about 50% of the total in the US.

Also congratulations for not caring about a 1 in 200 chance of death.

> Where's the data to back up any close to an assertion like this?

Latest IFR estimates are between 0.5% and 1.0% (1 in 200 to 1 in 100).

https://www.who.int/news-room/commentaries/detail/estimating...

Take a look at how mRNA vaccines work, and how they came to be. It is still mind-blowing to me.

Right, good point. It is a quality of uniqueness for searchability and marketability. On top of my head I can only think of a couple of counterexamples: Windows and Messenger, but your point is still very valid.

I guess in practice, for search purposes, we will see some kind of suffix attached to the name, Element Chat or similar.

I was trying to address the point of the parent comment, that using common words is detrimental to searchability. I think that is true, but as shown by the examples I gave, it is not catastrophic and can be worked around.

Processing: https://processing.org/

It is not great, but it is also not a disaster. See Go, React, Rust, Processing, Atom, Windows, etc...

I find useful to transform percentages to population scales. In your case, 99.4% recovery rate equals a 0.6% fatality rate. That translates to 1 fatality for every 166 people infected. Think about your social circle, how big is it? How many people there would be affected?

So about 1 person for every 400 infected. It also implies we are about 1/8 of the way from the maximum number of posible fatalities in the US.

Have you tried the Simplify GMail browser extension? It made the transition from Inbox to GMail less painful to me.

https://simpl.fyi/

Sure: https://www.technologyreview.com/2020/04/27/1000569/how-long...

One factor people often forget in the fatality equation is the attack rate, or how many people can be potentially be infected. For the flu, it is around 10% to 15% yearly. For C19, it could be >80% given that there is no natural immunity nor vaccines. Taken that into account, C19 could be at least 40x deadlier than the flul (Flu: 0.1% CFR 10% AR, C19: 1% CFR 40% AR). The 0.1% CFR for the flu is taken from the estimates of the 1957 and 1968 pandemics, which left 100K fatalities in the US. Estimates for the seasonal flu are lower than 0.1%.

Got it. Thanks. I think it may be because the IFR is less reliable, since for most diseases it requires a gross estimation the real number of infected persons. It is probably more popular these days because people are putting a lot of effort into the comparisons with the seasonal flu.

Thanks. I have seen serious estimates of 0.1% CFR for the 1957 and 1964 flu pandemics, but only notes saying that the seasonal flu CFR is <0.1%. Do you have a source for the 0.04% IFR?

Note that the immunity is not permanent. The average immunity to the endemic coronaviruses that cause the common cold lasts about a year.

No worries, I had made the same mistake. I just wanted to point out that the attack rate is also part of the fatality equation, and it is commonly not taken into account.