JackeJR

https://news.ycombinator.com/item?id=47676174

Dropbox is a lot more than file storage. The syncing itself has been through serious tests to characterise its behaviour. Sure, some may not like the decisions taken to direct its sync behaviour one way or another but at least all these are known through formal testing.

There was a discussion of a self-built dropbox on the frontpage (https://news.ycombinator.com/item?id=47673394). This is just to show that dropbox is thoroughly tested for all kinds of wierd interactions and behaviours across OS using a very formal testing framework.

Actually it can. See https://youtu.be/FUQwijSDzg8?si=LWd5gVNYRd3HH9rJ

Or just search for the James-Stein paradox.

For (1) you can prepay i think up to 10 years? And every year you just prepay 1 year again and you will have 10 years to remember that you forgot to pay a domain registration bill.

Well a good surveyor would take into account order effects, i.e. the order in which questions were asked for example using Latin square designs.

The thing is that with such a sample, we don't really know

1. If the effect is real. i.e. had the patient not been given the injection, would his/her condition improve spontaneously.

2. Assuming the effect is real, what are the circumstances that make the treatment work for this person.

Not to be overly dismissive of the good work but it is too early to be optimistic about this given the above and the fact that the results were not replicated out of Sana suggest that there is a lot that we need to work out before this becomes a viable treatment for the masses.

The harms of hyping this up is that readers will get their hopes up and then be disappointed when things don't pan out as do most scientific endeavours. Overtime, readers will learn to distrust anything that is being reported because 90% of which do not translate to real world impact. It is hard to get the nuance that "science takes many many failures and iterations" to the public and the more likely outcome of such reporting is general distrust of science when things don't go the way that is hoped for.

N=1 study should not have made it into headlines.

Although the research marks a milestone in the search for treatments of type 1 diabetes, it’s important to note that the study involved one one participant, who received a low dose of cells for a short period—not enough for the patient to no longer need to control their blood sugar with injected insulin. An editorial by the journal Nature also says that some independent research groups have failed in their efforts to confirm that Sana’s method provides edited cells with the ability to evade the immune system.

And also have parity built in for file recovery. The alternative will be to use par2 to create parity files.

Just have sane firewall rules and you are good. E.g. if I install openssh-server and it auto starts, it doesn't make it out of my machine because my nftables does not allow inbound on port 22. It's just knowing the default behaviour and adjusting your practices for it.

causal inference with GenAI

It swings both ways. In some circles, logistic regression is AI, in others, only AGI is AI.

There are many reasons this comparison is not made. I will just touch on one. The target medium is different. For html, you have monitors of different sizes as well as windows that can be resized. For latex, you choose your target at the start: A4 paper? Screen presentation? A0 poster?

With a fixed medium in mind, you can be extremely particular on where on this canvas you want a piece of text/graphic or whatever.

Without a fixed medium, you have to have logic to address the different mediums and compromises have to be made.

browser() ?

Granger's causality is a very restrictive and incomplete view of causality. Pearl's counterfactual system with do calculus is a more general way to think about causality. This SURD appears to be a souped up version of Granger.

I have not found an employer that values skills in multiple languages + quick at learning new ones. I assume these are the basic req for your type of work. Please share if you can how you got started .

I am quite curious about your approach, Do you have an example code somewhere?

Mindstorms is an example of what did not work. I want to provide an example of what does. BBC microbits. It has a visual programming interface that is translatable to python or JavaScript .

For this problem specifically, I have found that turning off auto mic volume helps to eliminate the "mute for first x secs if speaking".

The claim may just be a result of how cause of death is determined.

The screening target should be for malignant cancers, not all cancers.

Use this to get the functionality without setting up a windows server instance https://github.com/kimmknight/remoteapptool

I can't find a privacy policy and it's free so...

one reason is to make sure that the best and brightest are distributed amongst different industries.

Append to that "x years before I was born"

The researcher was studying dishonesty... So I gather that she was more interested in dishonesty?

Not sure I buy the interest is theft comment. Interest is paid for opportunity cost.

For the case of Syria, it is damned if you do and damned if you don't. Red lines were crossed (chemical weapons) and it was only when ISIS emerged that there were boots on the ground.

The recent updates have stabilised things a while lot with nice features like atomic updates.

I have seen chargers that simplify the logic to drop everytime a new device is connected or disconnected to recalculate the max power per port. So these days if I am shopping for a charger say for a IP cam, I assume that each usb C takes 100w and usb A to take 22.5w and calculate what the max draw could be. If the max advertise power budget is at or above this value, then it is a fairly good indication that there will not be any dropping of connection.

The chargers that drop connection is typically because their max power budget per port depends on how many ports are used. E.g if my max power budget is 200w and I have 2 usb c pd and 2 usb A. Each usb c pd can have a max of 100w per pd3.0 specs. So if I have 2 devices plugged in to both usb c, if I need to plug a USB A device in, I need to lower at least one of the usb C to 65W.