sporkland

12 ms·

61.
by sporkland 1y ago
yeah I've been thinking about them as stochastic content addressable memory. You can put as many next = userInput; while(true's) { next = mem[next]; } around them as you need in different forms. Single shot. Agents. etc and get
62.
by sporkland 1y ago
Over a decade ago I was starting a java process up with local MySQL and getting 45k rps read, request per thread with load wasn't hard to achieve. Not sure why this is an accomplishment.
63.
by sporkland 1y ago
I had a non-corporate personal PayPal account. I don't sell anything online, but someone sent me 50 bucks or something for some supposed product. Then they demanded a return from PayPal. PayPal took the 50 then charged me a 20 dollar f
64.
by sporkland 1y ago
Great succinct summary of a hard problem! I might reword: "attacker-controlled data, sensitive information, and a data exfiltration capability" to: "attacker-controlled data and privileged operations (e.g. sensitive informati
65.
by sporkland 1y ago
Having operated a large site with 1000's of services I've never had the metrics thread crash on a service. I've often seen the telemetry pipeline crash. If you've been writing the metrics to logs in a thread you at leas
66.
by sporkland 1y ago
> These crates also get "refactored" every few months, with breaking API changes, which breaks the stack for months at a time until everyone gets back in sync. This was a problem with early versions of Scala as well, exacerbate
67.
by sporkland 1y ago
This is one place where I hope the emerging LLM world helps protect us. If the automated filters actually use image or text models to decide whether it is frivolous or not seems like a tool we didn't previously have in our tool belt.
68.
by sporkland 1y ago
Everything is getting flagged that seems anti-currently elected's actions. I tried to raise it here, seems like HN for various reasons is over talking about this topic: https://news.ycombinator.com/item?id=43570025 I
69.
by sporkland 1y ago
Aren't humans and animals just prediction machines with in built goals around surviving and procreating? Plus it's a massive prediction machine trained on a corpus of the bulk of human knowledge. Feels weird to see it minimized in
70.
by sporkland 2y ago
Absolutely and unequivocally false. Unless you are casting Asian and Indian as white.
71.
by sporkland 2y ago
I honestly can't tell if you're referring to the current administration or the previous one with this comment. (Fun SNL video with similar confusion: https://youtu.be/8h_N80qKYOM )
72.
by sporkland 2y ago
Overclocking Celeron's those were the days. Intel binning down a bunch of processors capable of reaching higher clock rates but selling them as a lower end part was a boon for college students everywhere.
73.
by sporkland 2y ago
It is a somewhat though provoking question for me where you LLM's fall in automata theory? Are they non-deterministic finite automatons which would make regex's a cousin? Or does it require pushdown automata levels of capability
74.
by sporkland 2y ago
I remember playing that leaked first level on the tram with roommates in college over and over again. We were so psyched by it. All salivating for the final release. Next to playing Quake for the first time with 3d acceleration (3dfx vood
75.
by sporkland 2y ago
Gotta settle ownership in the octagon.
76.
by sporkland 2y ago
I look at stories like this and a key moment of failure that is obvious to anyone that has ever deployed code is don't make a change and roll it out to 100% of all devices/servers immediately. Feels like there is just some basic t
77.
by sporkland 2y ago
I like how you analyzed that, one addition: I also think some form of the anthropic principle is at play. The societies that exist have something that allows them to continue to exist. Free speech can allow a society to seek truth and being
78.
by sporkland 2y ago
This interview about dopamine and homeostasis in the body would support the idea that resetting that system through suffering would likely be a valid approach: https://youtu.be/R6xbXOp7wDA?si=iUUfMZe05DO0wR5K
79.
by sporkland 2y ago
I use https://theoldreader.com/ it's been reliable and trouble free since the death of google reader. I even use it to catch popular hacker news stories: hnrss.org/newest?points=150
80.
by sporkland 2y ago
Adjacent News Radio Marriage Proposal: https://www.youtube.com/watch?v=y-yGUSRdNG4
81.
by sporkland 2y ago
Reminds me of the Rich Hickey talk Speculation[1]. There is a special place in hell reserved for programmers that break back compat (for non-security impacting reasons) with widely used libraries, including google's guava developers.
82.
by sporkland 2y ago
Was this the plot of a silicon valley episode?
83.
by sporkland 2y ago
Agree. I'd say openid connect looks closer to SAML in terms of authenticating users and bootstrapping a "sessions" if you will. OAuth2 in my mind is one potential approach to maintaining a session, post initial authenticati
84.
by sporkland 2y ago
I remember them talking about self driving and the tesla's being so far ahead and then not being able to tell the difference between Cruise and Waymo. Waymo is so far ahead of everyone else as someone that uses them in SF it's no
85.
by sporkland 2y ago
Starla is crushed right now
86.
by sporkland 2y ago
Anyone have any takes on the Fitbit Ace LTE vs the devices listed on their approved devices list ( https://www.waituntil8th.org/devices )?
87.
by sporkland 2y ago
I've asked chatgpt to state its confidence after an answer and it's mostly said it's very confident, except onetime when the question was pretty ambiguous.
88.
by sporkland 2y ago
Do these occur without the use of SSRI's? e.g. in ADHD patients? What if _someone I know_ has had these brain zaps extremely infrequently but definitely noticed them in the past?
89.
by sporkland 2y ago
I hate how fetishists on certain narrow cross cutting topics (e.g. right to repair, open source, environmentalists, security) drag the conversation in such a negative direction. The biggest revolution in biking accessibility happened in th
90.
by sporkland 2y ago
I'm definitely of two minds on this topic: 1. I'm getting value out of the current batch of models in the form of accurate Q&A/summaries as well as tweaking or generating clear prose or even useful imagery well beyond wha

More ›