axutio

I've been playing Wordle as a saved PWA on iOS, wonder how the transfer will work for that.

> If you're applying for a mortgage, how do you expect them to verify your salary? You can submit paystubs, but those can be faked.

Why not keep this, and punish people who submit fake paystubs for committing fraud? I could also take advantage of my employer by faking other documents and signatures, but I (and most people out there) don't do that because that's illegal. If your potential employees are so untrustworthy that you can't trust them not to defraud you, you have bigger issues as a company anyways.

Statuspage: https://www.redditstatus.com/

Based on comments in third-party app communities, it seems like the issues are isolated to third-party clients and not the official Reddit app. Also interesting is that the issue was occurring earlier in the day, reported as resolved, and failure rates are spiking once again now.

Not arguing, but genuinely curious - what do kernels need randomness for?

This metric already exists in public health, in the forms of YLL (years life lost) and DALY (disability adjusted life years [lost]). In the case of something like COVID, YLL will simply integrate expected life span minus age at death across the population, while DALY will take into account the effects of lasting disability, for example counting each year of life as only 0.8 "disability adjusted life years" for someone who suffers chronic long COVID.

Because of the difficulty of applying objective disability effect proportions, however, YLL is a much easier metric to use and there's a lot more estimates available for it as well. I found an article estimating pandemic YLL in the U.S. over 2020 to be 7,362,555 (massive 95% PI though) [1]. Interestingly, they attribute only about three quarters of these to COVID itself and about one quarter to indirect effects of the pandemic, such as lockdown. Comparing against YLL values for leading causes of death from 2016 [2], the impact of COVID in 2020 was slightly less than that of ischemic heart disease (9,445,400 YLL) but comparable to the next four causes of death (lung cancer, RTAs, pulmonary disease, colo/rectal cancer) combined.

[1] https://doi.org/10.1371/journal.pone.0256835

[2] https://doi.org/10.1001/jama.2018.0158, Table 1

One interesting contributor to excess deaths is a spike in deaths attributed to Alzheimer's and other dementias - there were about 42,000 more of these deaths in the U.S. in 2020 than would be expected from the previous five-year average [1,2]. These deaths are primarily attributed to the negative effects of lockdowns on dementia patients - loss of access to family and caregivers as well as social isolation generally have a substantially more negative impact on this subset of individuals than on the general population.

Using a total 2020 U.S. excess death count of ~415,000 [3], this means that just above one in ten excess deaths can be attributed to the negative effects of lockdown rather than to COVID itself. On the other hand, the alternative would likely have been letting COVID ravage senior care facilities, which could have been much worse.

[1] https://www.aarp.org/health/brain-health/info-2021/alzheimer...

[2] https://www.alz.org/media/Documents/alzheimers-facts-and-fig...

[3] https://doi.org/10.1073/pnas.2101386118

This just becomes a question of pragmatism at that point - Google lacks the capacity to determine which of the blocked results are legal versus are not without incurring cost, so the most realistic approach is to recognize that a majority of results from that domain are illegal and block the domain. This is just the simplest way to enforce a particular rule in a particular case that can't otherwise be cheaply codified programatically.

From when I last tried it, Outline lacks any of the database/datatable functionality of Notion, but Appflowy seems to already have much of this functionality built in (or at least something like it is visible in one of the screenshots shown). Furthermore, logins in both the self-hosted and cloud versions of Outline require logging in with third party providers, no local accounts, so it can never be fully self-hosted.

I mean, I don't buy the conspiracy theory at all, but I would expect it to still be true that you benefit from DDoS attacks even if you offer protection for free. As the need for DDoS protection increases, you convert non-users to users, moving them one step closer to paying customers.

I'm sure that the fact that it's highly illegal and unethical are reason enough for Cloudflare to not sell DDoS capacity, but the perverse incentive is still there.

Also interesting is ISRIB [0,1], which is intended to negate the cognitive impacts of stress memory accumulated over years of immune response, and possibly as a result of TBI as well [2].

[0] https://en.wikipedia.org/wiki/ISRIB

[1] https://doi.org/10.1073/pnas.1815767116

[2] https://doi.org/10.1073/pnas.1707661114

Could you share your clone (w/ source code)? I've been wanting to build an open source Obsidian Publish alternative w/ the ability to authenticate and edit - Yjs + ProseMirror would allow for high quality collaborative Markdown editing on the frontend.

More broadly, I want a general Markdown notes "backend". Obsidian currently operates on local vaults which we can sync w/ general cloud providers, but no providers have all the features I'd want. Using Git lets me have great versioning of notes, but makes realtime syncing impossible, and using OneDrive is the opposite, so for the moment I'm using a Git repository in my OneDrive to get both.

It would be great to have a universal self-hostable note vault backend that interfaces with Obsidian, Dendron, etc. and allows for real syncing and publishing without requiring something hacky.

And when it fizzles out, it'll just become more fodder for climate change denialists down the line. Six years from now, they'll point to this as another piece of evidence that climate change isn't real/is a conspiracy/is heavily overblown. Alarmists have for decades and continue to harm actual progress in the fight against climate change.

The question isn't whether I choose to get my news from YouTube or not, it's whether YouTube presents itself as a legitimate source for news consumption or not - and it certainly does. The idea is that some large subset of YouTube viewers will, as a result, assume that they're being exposed to a basket of news representative of the breadth of opinion present among media outlets, when they're actually not.

> Youtube is an entertainment site, not a repository of legal filings

It might not be a repository of legal filings, but it's certainly not just an entertainment site. Opinion-oriented media, news programs, court footage, press releases, and educational videos have all been intentionally promoted in the past by Youtube.

The issue at hand is that some press releases are being permitted (and even promoted by YT) whereas others are not, despite the fact that there's no particularly notable difference in the authority of those speaking. (Obviously I recognize that this press release in particular more likely than not contains dangerous misinformation, but I still object to the idea that Youtube can unilaterally decide what is and is not dangerous misinformation).

I've been using DeepLabCut for physiology research, and it's really quite amazing. Manually labeling ~200 frames yields excellent results given decent contrast and lighting. Although the underlying technology itself is probably nothing new, it's packaged with a GUI that makes it very friendly to researchers across fields. The main challenge/obstacle, in my opinion, is hardware - training ResNets still requires hours of powerful GPU time, although DeepLabCut is also capable of using MobileNets with only a small decrease in performance.

I'm wondering how swapping Electron for another Chromium backend will allow them to halve the memory usage. Is Node the main reason that Electron uses so much memory? Or is the Chromium renderer poorly optimized to the extent that it's using double the memory needed?