Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
gojomo
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
47 ms
·
511.
▲
by
gojomo
4y ago
Not truly "unlimited digital copies" – just extra time-limited checkouts, as a temporary expansion of the prior 1:1 lending, and only while most of the country's libraries were closed. Those limits & circumstances are rel
512.
▲
by
gojomo
4y ago
Lots of laws were ignored or creatively re-interpreted, by plenty of people & entities, during the unprecedented pandemic health emergency of early 2020. Many emergency actions taken by elected leaders have been later found illegal by c
513.
▲
by
gojomo
4y ago
> Majestic promotes their list as the "top 1 million websites of the world" Well, the source URL provided by the article author initially claims, “The million domains we find with the most referring subnets”. Then it makes a
514.
▲
by
gojomo
4y ago
Many issues with this analysis, some others have already mentioned, including: • The 'domains' collected by the source, as those "with the most referring subnets", aren't necessarily 'websites' that now, o
515.
▲
by
gojomo
4y ago
Not all crypto transactions. Some blockchains effectively use advanced cryptography to obscure sender, recipient, & amount beyond any known methods of revealing them. Zcash & Monero are the leaders here. Zcash, used with its fully ‘
516.
▲
by
gojomo
4y ago
That image should be the 'Lena' of person-removal image benchmarks.
517.
▲
by
gojomo
4y ago
Thousands of years, or millions?
518.
▲
by
gojomo
4y ago
Cool idea! Reminds a bit of Jonathan Abrams' 'Nuzzel' - absorbed into Twitter Inc, perhaps now powering some features of 'Twitter Blue' subscription services. The auto-reload is a bit disorienting - I'd suggest
519.
▲
by
gojomo
4y ago
I used to think that way; especially in that every once-in-a-while, a flu strain comes through that affects older generations less, because it's like something they saw decades ago, but youngr people haven't yet. So, why not get e
520.
▲
by
gojomo
4y ago
The words "linked to" or "associated with" are tells that no real causality has been established – no matter how casually, or sloppily, causality is implied by the paper or breezy PR summaries. Maybe, early pre-diagnosis
521.
▲
by
gojomo
4y ago
Yes, COVID has since spread back to many animals, & viruses can travel far. But they still haven't found the precursors of the Wuhan outbreak's original strains. It is suspicious that EcoHealth lied about what research they ha
522.
▲
by
gojomo
4y ago
How would two different non-bat animals shed two different (but closely-related) bat viruses, in a market with no bats, and far away from the bat caves where related viruses are documented? And further, manage this with no evidence of any
523.
▲
by
gojomo
4y ago
When there are massive institutional incentives to proclaim X, it is only wise to discount, somewhat, proclamations of X. In Vanity Fair's lab-leak theory story, there was an illustrative detail: *> Inside the NIH, which funded such
524.
▲
by
gojomo
4y ago
A lab worker who was careless (if not depressed/malicious) could just as easily carry out two variants as one. One animal shedded... two variants of virus? How did the animal get it, when no other nearby wild or market animals in the (
525.
▲
by
gojomo
4y ago
The prior linked post is a gish-gallop of overconfident motivated reasoning. "Two lineages" in rapid-succession is weird - but no better explained by two zoonoses (that left no hints in any animal population) than two lab-leaks. W
526.
▲
by
gojomo
4y ago
Overfocusing on the market cluster is an error. First, because viruses travel easily – so even if 100% of early cases were around the market, it could have reached the market via a lab worker. And second, early overemphasis on the market cl
527.
▲
by
gojomo
4y ago
I don’t have a sense of the ‘normal’; it might vary by field. For example, for COVID vaccines, we’re being inundated with observational data - often informal, sometimes in reviewed studies - long after initial approval RCTs.
528.
▲
by
gojomo
4y ago
Such a review is necessarily limited to topics which have both an RCT and an observational study. As such, the general bias against publishing non-significant results could mean the conditions for being able to even try such a review risk
529.
▲
by
gojomo
4y ago
There are ways to hide IP endpoints from relaying nodes, as well.
530.
▲
by
gojomo
4y ago
Notably to some who may misinterpret the "stays in memory" aspect: even the copy in memory is fully scanned by every query. So it's not just disk access patterns that don't give anything away, but RAM access patterns
531.
▲
by
gojomo
4y ago
The server already has a full copy of (its subset of) Wikipedia. Every query touches every byte of that snapshot, in the same way... but the homomorphic-encryption math distills that down to just the fragment the client requested. And it
532.
▲
by
gojomo
4y ago
Interesting! But, it'd be helpful to clarify further the strength of the following claim: > This demo allows private access to 6 GB (~30%) of English Wikipedia. In theory, even if the server is malicious, it will be unable to learn
533.
▲
by
gojomo
4y ago
Of course, AirBnB misleads everyone by hiding key costs from sorted comparison views of the 'average nightly rate'. It is also clearly a company-wide value to mislead, as their subsidiary HotelTonight does the same thing with rega
534.
▲
by
gojomo
4y ago
Is a space-before-exclamation-point a Frenchism?
535.
▲
by
gojomo
4y ago
Apparently 'Amundi' is a French asset-manager, roughly Europe's BlackRock or Vanguard.
536.
▲
by
gojomo
4y ago
There are some industries, activities, teams, & individuals who can work as well, or better, remotely. There are other industries, activities, teams, and individuals who can work better with more in-person/same-site time. Some work
537.
▲
by
gojomo
4y ago
That's a fine micro-comparison to consider, alone! But also, what if this is a rare situation where the autonomous car completely fails compared to a human-in-seat? But, even this failure still resolves in under-a-minute. Or, worst-cas
538.
▲
by
gojomo
4y ago
As far as I can tell, none of the reported incidents in SF involved a "computer crash". Further, none have been as 'catastrophic' as other very-common accidents with human drivers. So yes, the details of recovery are imp
539.
▲
by
gojomo
4y ago
The cars relying on external routing guidance, from cloud services similar to 'Google Maps' driving directions. An alert from emergency services can simply... remove roads from consideration, the same as is already routinely done
540.
▲
by
gojomo
4y ago
But if the human-away-from-vehicle blocks things for minutes, while the temporarily-confused-AI has a remote-operator fix the issue in 30 seconds, or others in the area find another solution even faster than in the unattended-car scenario..
More ›