Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ckastner
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
29 ms
·
301.
▲
Learn and Relearn Your Field (2008)
(terrytao.wordpress.com)
1 points
by
ckastner
5y ago
|
0 comments
302.
▲
by
ckastner
5y ago
That is obviously absurd. There is no way The US, EU, China or any other nation will give up the power/capability of printing their own money. Especially so when giving up the one power would subject them to another that can be gamed
303.
▲
by
ckastner
5y ago
Sure, but that "second" you mention is an eternity compared to the max 30µs (see referenced paper) it takes a CPU to switch from highest to lowest frequency. Why would you expect critical temperature threshold shutdown latency to
304.
▲
by
ckastner
5y ago
This reminded me of the fascinating obversation of convergent evolution, the "independent evolution of similar features in species of different periods or epochs in time" [1]. For example, the concept of flight has evolved indepen
305.
▲
by
ckastner
5y ago
What is "quickly enough"? Isn't the reaction time today somewhere on the order of a few milliseconds? How much could a component overheat in that time? The maximum current, though high, is still limited, after all. Also, ther
306.
▲
by
ckastner
5y ago
> The EU and Germany in particular have decades of privacy regulation and core values behind them. So much this. This is evident, for example, from the very first sentence [1] of the GDPR: The protection of natural persons in relation
307.
▲
by
ckastner
5y ago
> cloud computing is often an easy sell for ARM and Amazon is likely going to be able to get the chips cheaper given that they're looking for markup on the service end while Intel needs markup on the chip. > Plus, I'd gue
308.
▲
by
ckastner
5y ago
Exactly. My favorite example are sites which require you to opt out of hundreds of third party processors individually (advertising partners who may receive data). That's as dark a pattern as it gets. It's also in clear violat
309.
▲
by
ckastner
5y ago
This is hilarious. So Craig Wright, in his continuing claim to be Satoshi Nakamoto (yet still refusing to prove it trivially), has -- naturally -- also claimed to own all of the Bitcoin that only Satoshi Nakamoto could have mined. Somebody
310.
▲
by
ckastner
5y ago
> I feel like there must have been something else since 2016? Funny, that was my feeling as well. I'm 99% sure to have read a discussion on this here, but all submissions after 2016 (apart from today's) have no comments at al
311.
▲
by
ckastner
5y ago
Indeed. Just to be clear, the internal complaint preceding the firing is apparently documented, and I did not mean to imply any particular motive. On the contrary: I found the that an otherwise unconnected third party, Walmart no less, woul
312.
▲
by
ckastner
5y ago
Whistleblowers becoming active after they have been fired are often ascribed retaliatory motives, but there's an interesting tidbit towards the middle: > Walmart in a 2019 lawsuit against Tesla said the latter's roof solar sys
313.
▲
by
ckastner
5y ago
It's a great read, but it got posted quite often here. Some of the previous posts have longer comment threads. https://hn.algolia.com/?q=how+sqlite+is+tested
314.
▲
by
ckastner
5y ago
Also interesting in this context: this recent paper [1] analyzed some of the most popular datasets and found numerous label errors in the test sets . For example, they estimated at least 6% of the samples in the ImageNet validation set wer
315.
▲
by
ckastner
5y ago
A good start would be to know why Google doesn't already penalize their rankings, as they have done with many others who have gamed their rankings. Yes, Pinterest does SEO like anyone else, but their ranking seems just outrageously ove
316.
▲
by
ckastner
5y ago
> By providing simple rules of information flow: small amount of content, retweet, followers) and rewards (likes), they've created the most efficient engines for finding and spreading novelty at the expense of everything else. Tho
317.
▲
by
ckastner
5y ago
Indeed, the title is misleading. However, I stuck with the original one because editorializing titles is frowned upon here.
318.
▲
How did Alan Kay read 20k books?
(quora.com)
6 points
by
ckastner
5y ago
|
2 comments
319.
▲
by
ckastner
5y ago
> Food poisoning is much more common than lightning. I said as much. The question is whether botulism is more common than lightning. Food poisoning is not uncommon, but can be caused by many pathogens, for example E. coli . Given th
320.
▲
by
ckastner
5y ago
Supposedly, flash freezing (which is distinct from your average home freezer unit) preserves quality such that even trained sushi chefs can't tell the difference between fresh and previously flash-frozen fish. https://www.ny
321.
▲
by
ckastner
5y ago
I am somewhat skeptical of this as well. It's not uncommon to hear of someone having eaten something "bad", but that can be the result of many pathogens and with regards to botulinum, the article gives a data point: > In
322.
▲
Attacking Hardware AES with DFA (2019)
(yifan.lu)
15 points
by
ckastner
5y ago
|
1 comments
323.
▲
by
ckastner
5y ago
The OP referenced CPUs and for at least AMD and Intel CPUs, firmware blobs need to be signed.
324.
▲
by
ckastner
5y ago
I guess that in their mind, they're producing code for some utopic x86 system that does not have non-open microcode baked in, in which case their approach would be right. This is a complete denial of reality, of course, and at the cost
325.
▲
Apple aims to build self-driving car with no steering wheel
(cultofmac.com)
2 points
by
ckastner
5y ago
|
2 comments
326.
▲
by
ckastner
5y ago
The pre-pandemic numbers are just as bad, though.
327.
▲
How Can an 11-Year Old Company Lose $844m on $661m in Revenues? WeWork Shows How
(wolfstreet.com)
32 points
by
ckastner
5y ago
|
11 comments
328.
▲
by
ckastner
5y ago
Nothing, probably, but the vast majority of the 500m MAU all but certainly use the official client, rather than a homebrewed one.
329.
▲
Cloudflare Announces Q3/2021 Financial Results; non-GAAP break-even
(cloudflare.net)
1 points
by
ckastner
5y ago
|
0 comments
330.
▲
by
ckastner
5y ago
The article references a similar fate experienced by the USS San Francisco [1]. I find seamounts somewhat fascinating; it's so odd that immensely huge features like these are still not yet charted. The Muirfield Seamount [2] is a good
More ›