ak47.io
tylerneely.com
messing with synthesizers and low-level systems in Berlin
192 posts
1,105 followers
211 following
Getting Started
Active Commenter
comment in response to
post
goes so hard I might now need to be evaluated for respiratory irritation
comment in response to
post
Die Idee ist, dass der Zielort die Pflicht hat, es direkt zurückzuschicken, aber außerhalb der EU ist das eindeutig nicht so, wie es funktioniert.
comment in response to
post
The advent of germ theory, vaccines, sanitation, and antibiotics, in this view, indirectly lead to an increase in social services, and a decrease in capital punishment, murder, slavery, and torture. In effect, it raised the value of human life in human eyes.
comment in response to
post
I’m pretty happy to know that my name is publicly associated with buying drones to blow up invaders. I really liked visiting St. Petersburg in the past but times change and values matter. I would feel bad if I didn’t get imprisoned the moment I stepped foot there lol
comment in response to
post
I'll kick off claude sketches of ideas from my phone sometimes but having at least a laptop screen to work with and commit bigger changes is necessary for me.
comment in response to
post
Important to keep in mind you can't get the first dose of the MMR vaccine until you're a year old, and in the US you don't get the second dose until you're around 5. As of today, 157 cases in the measles outbreak are in kids under 5 years old. So 33% are simply too young to be fully vaccinated yet.
comment in response to
post
Why not host it somewhere low risk for world leading cryptographers to actually attend?
comment in response to
post
Your assumptions of pi, gravity, units of time, institutions, borders, the utility or obtainability of truth, basically there are no correct theories about reality and it’s not possible or useful to obtain them anyway
comment in response to
post
The fact that applying Girard’s earlier work on linear logic basically led to Rust’s compile time guarantees makes me curious what else will be possible to cheaply exclude at compile time with other light twists in semantics. I wonder what it would mean to translate the above result into a PL
comment in response to
post
This reminds me of a suggestion I once heard that new logics like Geometry of Interaction might lead to programming languages that let you relatively cheaply understand if certain invalid states could be reached at compile time, but I haven’t had a chance to understand the real limits yet
comment in response to
post
When there are others, it’s definitely important to coordinate on the interfaces and module contracts, and do the required communication to get others on board with seeing a lot more red diffs that might remove things they thought were proud of. It can be an exercise in emotional detachment lol
comment in response to
post
Damn, Habeck too www.reuters.com/world/europe...
comment in response to
post
Also Scholz has excused himself: www.tagesschau.de/newsticker/l...
comment in response to
post
I hope not but tbh I don’t have much familiarity with the Wahlkreise where they draw support
comment in response to
post
If the choices are CDU+SPD+Grüne, CDU+SPD+FDP, or CDU+AfD, the internal politics w/ Söder will be crazy as he views the greens like we view the AfD, and whether Scholz will tolerate being in another FDP Koalition is questionable
comment in response to
post
While they probably won’t make the 5% hurdle, they could still meet the 3 Erststimme candidate hurdle as Die Linke did in 2021. If they both make it, that dilutes the CDU+SPD‘s proportion and may prevent a simple majority with them, and that’s where CDU+AfD starts to seem a little more possible
comment in response to
post
Non-paywalled link?
comment in response to
post
It looks really gross until you fry the hell out of it and then it’s basically pretty similar to black pudding! I cringed when I first tried it but actually kind of enjoy it now
comment in response to
post
People everywhere are often kinds of assumptuous when information flows in the opposite direction of a familiarity gradient.
comment in response to
post
So much, maybe even most food is pretty regional. My dad is obsessed with Cincinnati Goetta, which most Americans would not know is American, although it has roots in a German food called Pinkelwürst that is not well known outside of northern Germany. (Also amusingly close to the word for piss)
comment in response to
post
They will continue focusing on this kind of thing until it's utterly beyond the human capacity to willfully ignore the uncomfortable truth that we need to come together and defend ourselves, and that some of us will actually need to do that actively.
comment in response to
post
While I think the increasingly harsh trade-offs of nonlinear carbon abatement curves in the face of growth-oriented economies will eventually cause carbon-fixated approaches to give way to more brutal degrowth models, practicing the act of collective action-oriented model making is crucial now.
comment in response to
post
The best thing you can ever do in the face of bad news is to roll up your sleeves, meet with others, and oppose what must be opposed. Together you'll find meaning and often friendship. Regardless of immediate impact, that network will be valuable as future information emerges that demands action.
comment in response to
post
I think I‘ll do that for the first time today, too!
comment in response to
post
Thanks Nick ❤️ I’m still very much in the afterglow. Although in a way it’s also a loss of an enemy windmill that I had partially structured my life to defeat. But until another one appears, it’s nice to have my hardest decision basically boil down to how much time away from work do I want to take 😅
comment in response to
post
Here's an example of ChatGPT doing something interesting with a normal programming language: sha512 hashing "hello" 507 times, which it would not have correctly predicted using a plain LLM. We'll see more multi-modal stuff like this over time using SMT solvers for cooler stuff etc...
comment in response to
post
ChatGPT 4 can write and execute its own code in normal programming languages. If you mean pure LLMs then it's something like O(N^2) on the number of tokens.
comment in response to
post
The thing that appeals to me about the heap-backed LSM block approach is that it gives you a way to decide at runtime how much effort you will spend compacting overlapping ranges in order to make the overall block index more specific, while avoiding copying unchanging blocks as a plain LSM would.
comment in response to
post
LSM, Bε, Bw trees all try to avoid moving data around until doing so can be amortized, with them forming a spectrum of how specifically a particular block/page can be rewritten in response to write locality. Plain LSM is locality blind. Bε rewrites data as it moves back. Bw is hyper local.
comment in response to
post
Compared to a Bw-Tree where you need to maintain pointers to every delta update that needs to be read along with the base page for a point lookup, this has a much lower metadata:data ratio, but that metadata is less specific, so filters are useful.
comment in response to
post
And one of the nice things is that it is fairly clean to bolt this onto a heap-backed B+tree to give it nice LSM-style asymptotics for random write workloads while still retaining optimal B+tree write amplification for sequential writes via the block gifting trick.
comment in response to
post
With an LSM, you tend to want to maintain an index of the ranges of keys that each sst block holds anyway. And for this block-first style, it does the same. But compaction becomes a bit more sophisticated, where range density can be more explicitly reasoned about.
comment in response to
post
While a naive Bε-tree still has the suboptimal LSM behavior on sequential writes of copying data several times from the tip to the leaf, this block-based LSM lets sequential writes immediately be "gifted" to the last "level" (but levels no longer exist as such).
comment in response to
post
it's also 20x more expensive than the GET that you'd do with at least a high fidelity cache of latest successful root metadata write.
comment in response to
post
Thanks :) Try the test online, as they publish the actual questions, so you can practice it beforehand: www.einbuergerungstest-online.de
You only need 17 out of 33 correct answers. I found it to be very easy.
More difficult might be the B1 language test, depending on how much practice you get.
comment in response to
post
On the other side - by the time you reach 31.7 years old, you have already been alive for one billion seconds. It's so much more than anything I can honestly comprehend.
comment in response to
post
That's gorgeous - especially as the mercury drops, Beef Wellington seems like a fun project to attempt
comment in response to
post
The other source of information that is extremely thorough but also extremely dense, and contains all of the guidelines for how the Landesamt für Einwanderung handles all immigration cases, not just citizenship (citizenship in section "S") is the LEA Verfahrenshinweise www.berlin.de/einwanderung...
comment in response to
post
That said, it may not actually speed anything up if it's incomplete, since it will end up in somebody's backlog that they may only revisit once every 3 months or something. But maybe it does in some cases.
comment in response to
post
I have heard from some people who applied more recently that there might be some stronger requirements about beginning the process only after you get certain pieces completed, but there is a chance that those people were confusing the "quick check" document quiz with the actual application.
comment in response to
post
It was a total horror show before 2024, but now Berlin has one of the fastest systems in Germany with the opening of the new citizenship processing center at the Landesamt für Einwanderung. The FB group is actually a pretty good source of up-to-date info: www.facebook.com/groups/13314...
comment in response to
post
I tried to begin my process in November of 2022, but the now-defunct Staatsangehörigkeitsbehörde Neukölln told me to go away because they had too many applications lol... But when Berlin opened the new electronic system this year, I submitted my application right away. But without the test done yet.
comment in response to
post
There has never been a better time! It's pretty common for it to take between 3-6 months in Berlin, if your application is submitted with all requirements already met. I'd try to get the Einbürgerungstest done ASAP if you haven't already, as that takes months to grade. I did mine in Potsdam.