Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
mjb
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
16 ms
·
121.
▲
by
mjb
3y ago
Metallurgy is super interesting. The fact that metals - these crystals that would be super brittle and stiff except for a few dislocations - can creep like this is amazing. My (slightly) tongue-in-cheek proposal for Arecibo: drones. Instead
122.
▲
by
mjb
3y ago
"Among the Thugs" by Bill Buford, which looks at football violence and crowd deaths in the 1990s is a great read if you're interested in crowds (and even if you're not interested in soccer/football at all). It'
123.
▲
by
mjb
3y ago
If you're interested in learning more about that, check out our NSDI'20 paper on how we chose this direction ( https://www.usenix.org/conference/nsdi20/presentation/agache ) and the Firecracker source
124.
▲
by
mjb
3y ago
Round robin can work very well when the variance of task sizes is small, but degrades quickly when the variance of task sizes becomes large (i.e. there are some tasks that keep servers busy for minutes, and some for milliseconds). In distri
125.
▲
by
mjb
3y ago
See my reply here: https://news.ycombinator.com/item?id=37175685 Short version: it's robust to staleness and concurrency, which is important in distributed settings ("find the best" is hard to do when "b
126.
▲
by
mjb
3y ago
Two random choices has a few significant advantages over "find the least filled one": - It's O(1), rather than O(buckets). - It's robust to stale data and concurrency. One way systems try to avoid the O(buckets) work is
127.
▲
by
mjb
3y ago
This is really great, actionable, sensible, advice for folks who write on the internet. It agrees with my experiences, too. On the other hand, it's sad that the culture we have built on the internet forces talented people like Julia to
128.
▲
by
mjb
3y ago
There's a terrible/great genre thriller by Jeffery Deaver called 'The Blue Nowhere'. Whoever ghost wrote it seems to have set themselves the goal of using every word from the jargon file. Unintentionally hilarious.
129.
▲
by
mjb
3y ago
All distributed locking systems have a liveness problem: what should you do when a participant fails? You can block forever, which is always correct but not super helpful. You can assume after some time that the process is broken, which pre
130.
▲
by
mjb
3y ago
> daily occurrence when you're operating at S3 scale Yeah! With S3 averaging over 100M requests per second, 1 in a billion happens every ten seconds. And it's not just S3. For example, for Prime Day 2022, DynamoDB peaked at ove
131.
▲
by
mjb
3y ago
That was a good read. However, it seems like the section on geo engineering does exactly what the rest tries to warn against: predicting behavior using simplified first order models. I'm biased towards believing the conclusion the auth
132.
▲
by
mjb
3y ago
The biggest problem with adaptive rate limiting is that, in many contexts, clients value predictability more than they do peak performance. Adaptive rate limits change client behavior in ways that clients can't predict or control. Even
133.
▲
by
mjb
3y ago
You can (sort of) turn thin dried spaghetti into ramen by presoaking it in a strong solution of sodium carbonate.
134.
▲
by
mjb
3y ago
It really is a zoo (and the whole of Western Washington is generally). If you're interested in the topic, I really enjoyed https://www.amazon.com/Roadside-Geology-Washington-Marli-Mil...
135.
▲
by
mjb
3y ago
Plate techtonics was very new (and rather controversial, as I understand it), science in the late 1960s: https://en.wikipedia.org/wiki/Plate_tectonics#History_of_the...
136.
▲
by
mjb
3y ago
Related, there's a cool transition if you hike up the Greenwater river (starting at the town of Greenwater, heading towards FS70) there's a cool transition where the geology switches from Osceola mud to the more typical Cascade fo
137.
▲
by
mjb
3y ago
For comparison, this is between 4 and 5 times larger than the Osceola Mudflow ( https://en.wikipedia.org/wiki/Osceola_Mudflow ) that formed the 550km² of fertile plains around the south end of Puget Sound (near Seattle)
138.
▲
by
mjb
3y ago
McMansion Hell isn't about taking the piss out of one's superiors, it's about looking down on the lower classes without the refined architectural taste of the author and audience. It's punching down.
139.
▲
by
mjb
3y ago
I viscerally dislike McMansion Hell. It's so mean-spirited and snobbish. People build themselves homes in the suburbs for their families to live comfortably and safely, and some jerks on the internet act all superior about how they hav
140.
▲
by
mjb
3y ago
> This is why mathematical libraries will have not only gamma functions but also loggamma functions. This is super useful for combinatorial problems too, because gamma(N+1) = N!. Some examples where I've used it include: - Erasure c
141.
▲
Bélády's Anomaly Doesn't Happen Often
(brooker.co.za)
2 points
by
mjb
3y ago
|
0 comments
142.
▲
by
mjb
3y ago
They did: http://web.archive.org/web/20200814020351/https://www.oceang... (At least twice, and probably more times).
143.
▲
by
mjb
3y ago
They did: http://web.archive.org/web/20200814020351/https://www.oceang...
144.
▲
by
mjb
3y ago
I did mean "antidote", thanks for the correction. I did read the book, although that was around the time it came out and my memory may be faulty. I remember not being particularly impressed with the level of evidence presented, an
145.
▲
by
mjb
3y ago
They aren't just daredevils, though. Check out "Heuristic Traps in Recreational Avalanche Accidents" http://www.sunrockice.com/docs/Heuristic%20traps%20IM%202004... for one interesting example: a lot of
146.
▲
by
mjb
3y ago
Is it a rare example, though? Taleb and others spread this idea that "skin in the game" is some kind of magical anecdote to faulty thinking, incomplete risk analysis, incorrect application of heuristics, and other things. That jus
147.
▲
by
mjb
3y ago
There are relatively few things I've read that I can definitively point to as having almost instantly changed my mind on something. This is one of them. Reading often offers a broader understanding or different perspective, but seldom
148.
▲
by
mjb
3y ago
It is cool, but that wikipedia article sure has an axe to grind. I'm not sure I'd agree with the "simpler" or the assertion that CH needs to store tokens. And the variant of rendezvous hashing that doesn't require O
149.
▲
by
mjb
3y ago
> How do you build a system around a database where values can no longer be valid at any time? This is a difficult problem, and there aren't nice general solutions. If you read the original Amazon Dynamo paper, for example, there&#x
150.
▲
by
mjb
3y ago
This is a good troll, and makes some good points, as Andy Pavlo always does. I'd diagnose the core problem differently, however. What application and system developers care about in modern architectures is application level correctness
More ›