breatheoften

How is it legal for it _not_ to be? Keeping that information hidden drastically incentivizes naked short selling -- and there is an exploit for naked shorts that will be incredibly difficult to prevent going forward ... no -- the exploit does not really rely on collective action -- it relies on the perception of collective action -- which is definitely fake able in an automated way with current technologies.

Naked shorts at this scale are gonna have to go away and no one will have more incentive to reach that goal than main street wall street ... ideally they can and should do this in a way that simply involves making more information public ...

No comments on this? This is the most interesting description of the phenomenon I've seen yet! Really nice writing.

Is there a way to know the amount of naked shorts that were sold? Wouldn't most folks (including market makers) agree that Naked shorts are obviously evil?

As the scenario likelihood decreases, the probability of correctly identifying the associated recovery procedure probably also tends to decrease -- particularly for cases where recovery is not super clear and for cases where the scenario itself is memory intensive to describe or baroque -- "a occurred before b instead of b before a as normally expected and sensor d has condition c..."

I wonder if simulated error scenarios ever occur during flight -- could be a good way to expose pilots to rarely used recovery procedures within realistic day to day contexts ...

She should never have been put in jail and each day she remains increases the injustice.

US society is absolutely unhinged with its extremely assertive willingness to punish the most vulnerable while simultaneously bending over backwards to avoid accountability for anyone with money or belonging to the correct political clubs ...

Interesting -- i wonder what kind of tricks evolved to better arrange windows? Zooming around without knowing "the lay of the window landscape" sounds like it could get tricky as you expand the amount of window state to a very wide range.

I could imagine this kind of system might necessitate invention of actual good way of auto-rearrange-all-windows strategies?

I don't see windows themselves as a bad thing -- it's having to arrange them that I don't like and wish the os could define conventions for window placement to allow the system to "do what i want" more often without me having to fight it by moving a window around...

The context as I see it -- one of the largest (if not the largest) social media megaphones controlled by a single human was recently silenced because of a recognition that this megaphone was being actively used to cause widespread harm.

The eff response to these actions is to emphasize that such blocking should be allowed so long as a consistent set of rules is used to perform blocking for this reason across a whole network. And they suggest the communication scale of the actor being blocked should not be a factor in the management criteria applied to the speech flowing across a given channel.

That's a principled position but it's also ludicrous - and it's impossible to take advice from that position which can then be applied to the real world. It's as tho they've said "yes, we acknowledge that free speech faces limits -- such as no yelling fire in a crowded theater -- but we do not acknowledge the existence of crowded theaters". What kind of advice are we left to take from such philosophizing?

Text sent over the largest of megaphones is just not the same thing as text sent over the much smaller megaphones available to most social media users.

Taking the EFF position espoused here to an extreme for illustration -- suppose Alice decides to use the emergency broadcast system as her personal twitter account for a day. With her every message all the cell phones in the country buzz an emergency broadcast alert letting us know what Alice wants to say. She exercises her right to political speech to let us all know that she thinks birth control pills are evil -- and in so doing she consumes some seconds of attention from 300 million people. At the same time -- she cross-posts to her twitter where she is followed by 50 people.

The EFF are more or less suggesting that the same principles and rules should apply to her twitter posts and her emergency broadcast system posts.

But in reality -- do we consider these information publishing channels directly comparable? Are we forced to accept that because these are both information publishing channels -- the set of things Alice is allowed to say on both channels must be the same?

Obviously we are not forced to take this position. And obviously this is a bit of a silly example since we all already know that different rules apply to the emergency broadcasting network -- yet we don't use the word "censor" to describe the set of things we don't want Alice to send out over the emergency broadcast network -- and why is that?

Going back to my claim that they are silencing themselves -- I think the EFF position in this article is basically abstract philosophy and that they are going out of their way to avoid addressing any of the specifics required to guide action in the real world. They had an opportunity to say something useful -- instead they said nothing in more words.

I think EFF should look for improvements to their communication on this topic. They need to be providing much more practical leadership guidance.

The ideological goals don't need compromise -- but the practical reality is that our society _needs_ an organization like EFF to be at the forefront on these topics -- advocating for better abstractions to help shape how we as a society should think and deal with these complex issues. EFF shouldn't be effectively silencing itself by becoming a voice that just repeats platitudes. This article to me felt like the right wing response to gun violence events "we send our thoughts and prayers but let's not politicize this tragedy".

In particular -- there are simple ways to avoid worry over the "slippery slope" associated with defining policies that apply different rules to different users -- you can espouse policy frameworks that are explicitly parameterized by impact (impact is a bit fuzzy to capture in truth but conceptually clear -- and simple proxies suffice to communicate the concept -- in most cases the number of followers is a reasonably close conceptual proxy for a social network).

If you have 70 million followers and you say "let's punish immigrants" it's just simply a different thing than if you say the same thing and you have 10 followers -- pretending there "must not exist" a different framework for responding to events that occur on such vastly different scales is not practical and it's actually a failure to provide good advice here on an ideological level -- because we _know_ that these events are different and project meaningfully different risk profiles onto the world.

There is not a clear ideological reason to falsely pretend that "the same thing" is happening when two users with such vastly different impact publish the exact same sequence of words ... The threat model for "danger of communication" is based on impact and so should be the policy frameworks aimed at managing these risks.

A platform is demonstrably different than speech in real-world terms. It is simply not possible to 'say' something while you are sitting on the toilet to 70 million people within a few seconds.

The idea that 'a platform' and 'speech' must be treated the same is ultimately a kind of class warfare -- its a false comparison that serves to benefit the rich (in this case, rich in terms of platform reach) by pretending that any attempts to place restrictions on the use of mega-platforms must necessarily imply new restrictions on the speech of the vast majority of people without such reach.

I'd expect any python codebase which tries to use async/await based co-operative multi-tasking for concurrency at a larger scale would inevitably run into long-tail latency walls -- not just once but repeatedly as the codebase grows/changes.

The potential for a code change or addition to produce unexpected synchronous blocking (potentially with GIL held) _somewhere_ seems really high and I have to imagine it would be very hard to avoid in the context of a real application codebase ...

I clarified in my other comment -- I wasn't trying to imply that the violence here was a consequence of the extremists involved having any legitimate claim that their free speech has been curtailed.

I think this riot is best understood as being actively organized by the current ruling party.

I think there is a clear free speech issue in the current politics though -- systematically devaluing the potential for productive political talk is a form of free speech restriction -- it's a ddos attack against rational discourse - which has the effect of reducing the value of political discourse in general.

Sorry I didn't mean to imply that the anger component here was related to any specific party complaining about loss of political speech in a disingenuous way -- I was more attempting to diagnose the current overall bad state of the politics as being a result of the way that pressure is put onto "freeness" of speech when an authoritarian regime actively creates an engine that makes reasonable discourse difficult or impossible ...

Maybe the problem is more an unequal distribution of "freeness" of speech across the society; grouped as the ruling political class, the opposition political class, and the distribution of thought in the populace.

The ruling class says whatever it wants with belief that there can never be any negative results for any action they take (max "freeness"),

the opposition can't talk about anything in a substantive way because all they do is react to the inane and random political grenades thrown by the ruling party (very constrained "freeness" really. Being forced to respond to propaganda-maximizing controversy after propaganda-effectiveness-maximizing controversy is a record that sucks to play and is definitely forced onto the air more than everyone wants -- and the targets of the propaganda blame the wrong people for why they have to keep listening to it ... )

the people are left with no meaningful political voice (0 freeness of speech) because there's virtually 0 correlation between anything being talked about in the political dances and anything that is actually sufficiently practical to talk about as to be worth the cognitive attention required to talk about it ... You can't have free political speech when there are no political engagements worth talking about ...

The "team sport" that is the current political landscape is not at all a fun or useful game -- some amount of fun and usefulness is gonna need to be found and introduced to the process of defining government to help move out of this ...

> Yes, those who test positive should have greater options for self-isolation than they currently do.

Nothing along these lines even remotely happened. The boogey man of 'forcing people to quarantine' can be avoided while still creating incentives and expectations that individuals should opt-into a plan that concentrates risk and concentrates isolation requirements around known risks.

A positive test in the West comes with vague advice to quarantine at home. It should've instead come with a pre-paid package providing hotel+food+income-supplement that continued for the quarantine period so long as you don't violate the isolation expectations ... Incentives, not punitive measures, would've sufficed to make it possible, easy, and expected for people to _avoid_ spreading to their co-dwellers/neighbors or anyone they come into contact with by necessity for basic survival needs.

I think there might be some truth to this ... observability of the pandemic is very highly non-uniform -- even assuming the input measurements can be somehow normalized in a semi-trustable way (probably not true in general), non-uniform population distribution + exponential makes for much much more noisy measurement than I think is regularly acknowledged.

I feel like there's a kind of built in bias against fully appreciating the effects of the unknowns given the nature of the crisis -- because being the guy who says 'well actually we might not _know_ that' in every policy discussion about the pandemic doesn't actually contribute much benefit ...

What is buying khakis a reference to? Seems oddly specific

I'm looking for a solution to the problem of exposing a port on my laptop port to the internet (let's say for example purposes: ssh server port) from whatever network my laptop happens to be connected to ... I want to be able to initiate connections back to my laptop from random devices which are connected to the internet via network infrastructure that I don't control ...

Can tailscale give me that without me having to setup my own network infra somewhere?

I loved these two articles and feel some amount of trust for the author's company based on the communication style and knowledge sharing -- just wondering before I dive in whether this tech offers a good solution to my problem?

I think their cloud play _will_ be B2C driven.

I predict they will build customer-facing 'cloud-powered features' where 'third party app code' can run on the apple cloud infrastructure 'on behalf' of a particular apple customer.

They will present a utility computing model to customers -- where each customer sees 'cloud-computing usage charges' displayed by app.

Ultimately, the cloud-compute/cloud-storage charges will tie back to the customer associated usage costs for storage or some other abstract usage based model of for cloud resource costs.

"Apple Cloud: build features for apple customers using usage-metered cloud resources -- with the customer paying for the usage charges they incur."

The article makes the claim that models which show similar train and test losses demonstrate minimal overfitting -- and are therefore less likely generally to exhibit a lot of text memorization.

I wonder the degree to which this inference is true in practice with respect to information like phone numbers... How exactly are the train and test sets formed in a de-correlated-with-respect-to-memorization-of-phone-numbers manner for models of GPT class that are trained on corpus's the size of the internet?

If a particular person's phone number occurs 1000 times in the corpus prior to being split into train/test sets, what are the chances that the number only appears in either the train or test set but not both?

deno seems really amazing. This is essentially single file distribution of a secure runtime for application code -- that is a lot of platform-capability-bang for the distribution-reach-complexity buck! This has to already be the best server container format for typical application code in terms of the security possibilities doesn't it?

I'd almost like to see deno grow some kind of puppeteer based browser api as server-app platform --

http request -> deno server process runtime (with secure sandbox) and puppeteer-like handle to the client browser for state transfer -> client ui render ... it would be quite interesting to think of the client browser page tabs as a "child process launched by the server" rather than as as a stateless request from an http client -- as most server rest api architectures tend to push you towards ...

> My concern is only will users accept a local server running on their desktop.

This sounds pretty cool -- is there a way for the deno runtime to restrict the server process to only be able to generate the legal types of http traffic that a browser could generate if allowed to open a connection to a specific url (or set of urls)?

If so i'd feel more comfortable about running the embedded server -- ideally the security risk of running the server would be closer to the risk when running an embedded set of daemon browser tabs -- rather than the risk when exposing all the unfettered power of a unix process ...

paraphrasing the antagonists digital revolution sales pitch

"We'll add click analytics to government websites to digitize the flow of information and solve the pandemic!"

Even linear regression is not very easily explainable for real world systems with currently utilized technologies -- especially when you have a virtually unlimited range of mechanisms intentionally or unintentionally subset or modify the sampling process (or select for desirable results at "sampling design time") ...

This is pretty dystopian ...

1. There exist laws to prevent discrimination against people based on protected attributes 2. ML models make predictions based on attributes without interpretability (it's not possible to prove that protected attributes are not factoring into model predictions) 3. Empirical observation that a model proxies a protected property exposes corporation to liability for regulatory non compliance 4. Therefore any study that could expose bias of a model used in production is to be road blocked or prevented ...

To combat flows like above -- seems like regulators are going to need to update rules with third party audits and an incentive structure that encourages self-regulation and derisks self-detection and self-reporting of non-intentional violations... ideally google should not be put into a position where it is incentivized to police its own ai ethics research to ensure that such research doesn't expose their own illegal/non-compliant activity ...

I'm not sure I really think of function coloring as a "problem" ...

facebook is experimenting with auto differentiation for Kotlin and looks like it's adding a new "differentiable" function color -- https://ai.facebook.com/blog/paving-the-way-for-software-20-...

It looks very easy to reason about and use to me ... and i personally find async a similarly useful marker ... It's about being able to push constraints from caller arbitrarily far down the callee stack -- which is really not something that types support at all but provides for a very high confidence variety of constraint -- and high confidence constraints seem to me like they convey a ton of information.

I've been wondering actually whether "function colors" might actually just be a good way to create a whole variety of strong statically enforceable constraints for functions. It seems like they lead to very good and simple programmer mental models ...

Are there languages that offer "user definable" function colors? I can think of a lot of application domains that would be much better served by these kinds of constraints than oo or other type-centric approaches ... it would be ridiculously useful to be able to mark a function with the "MyDomainBusinessLogic" color and get assurances that such a method can only call other functions annotated with that color ... would provide an easy way to iterate on app specific abstractions provide compiler assistance for the communication of layering intent -- rather than a bunch of poorly specified words in documents that try to communicate layering intent to other developers -- in language that is either sufficiently precise as to be incomprehensible -- or sufficiently vague as to be subject to (mis)interpretation ...

Why do these parties not agree on the timeline for the review period? One party claims they had been going through review for 2 months and one said the review was submitted with only 1 day ...

It seems likely that both sides are mis-characterizing events in pursuit of their own goals -- which to me suggests a general environment where parties are tending not to communicate with the goal of arriving at agreement -- but rather as adversaries.

Hackernews will of course assume that the only reason for an adversarial relationship in this scenario is "cancel culture SJW memes" and other "deemed toxic" trends for which the minority voice is almost always the only guilty party.

I think that level of dismissal is unwarranted as it seems pretty likely to me that there is a definite gotcha-oriented communication style coming from the google leadership here -- which very well might reflect an intentionally or unconsciously adopted strategy to create obstacles for voices of disagreement like this.

Is there a difference when collecting unemployment when fired vs resigning?

Another risk factor that I doubt anyone is considering -- if the vaccine is rolled out and ends up not being effective (if vaccinated populations still die from covid at more or less same rate once the vaccinated population becomes sufficiently large) -- then this could lead to a real, long term, and very unfortunate surge in anti-vaxxers ability to persuade people not to take good vaccines that are proven to be effective ...

At the end of the day, a Smallpox pandemic in 10-20 years might end up killing more people than a premature covid vaccine saves ... No idea how you could quantify this risk though -- but seems unfortunate that medical science has no way to realistically factor in this kind of third order effect into their risk modeling ...

From what I've understood -- it is not yet known whether having the vaccine will prevent or reduce transmission.

I believe that most experts are expecting it won't.

I wonder the degree to which this is about apple trying to reduce their services crypto workload? When developers say they "want a mac instance" they often "want a mac instance that can talk to apple services like a normal mac instance" in order to sign apps and such. I'm not sure but I think arranging for that capability probably involves round trips with apple services to get required key signing artifacts into place ...

If instances are totally ephemeral in the same way as we are used to with linux vm's maybe that runs the risk of creating a real problem for apple services?

I've been wondering what kind of hardware optimizations might be possible based on tagged pointers? If you know a memory location is uniquely referenced for example when you load or store ...