@JMRothberg developing a precise optical system on a boat has to be the dumbest part of this whole post
English
Maniac Selinsky
582 posts
Maniac Selinsky
@ManiacSelinsky
increasing sacrifice
Katılım Haziran 2022
810 Takip Edilen102 Takipçiler
Project Summary – Open-Source Low-Cost DNA Sequencer 🧬
We are developing a fully open-source DNA sequencing platform designed to dramatically lower the cost of sequencing for small labs, education, and hobbyists.
Target specifications:
•<$500 instrument cost
•<$50 per sequencing run
•~200 million bases per run (low throughput, high accessibility)
Technical approach:
•4-color sequencing-by-synthesis
•Surface-based DNA amplification (cluster generation)
•Focus on robust, simplified chemistry that can run on low-cost optics and hardware
•Designed for large, bright clusters and relaxed imaging requirements (e.g., Raspberry Pi–class cameras)
What makes this different:
•Prioritizes cost, simplicity, and accessibility over maximum throughput
•Built from the ground up for open-source replication
•Integrates chemistry + hardware co-design, not just adapting existing systems
Current stage:
•Early-stage development with defined architecture
•Need to implement and test sequencing chemistry and surface amplification workflows
•Parallel development of low-cost imaging and fluidics
Team & environment:
•Led by a high school founders working with experienced mentors
•Hands-on, rapid iteration environment
•Summer work conducted on a research yacht in the Mediterranean & Aegean Sea
Who we’re looking for:
•Strong experimentalist with hands-on experience in SBS chemistry and surface DNA amplification
•Able to build, troubleshoot, and iterate quickly
•Interested in pushing sequencing towards ubiquitous personal access.
English
Sick and tired of nefarious timeline manipulation or demonic incursions?
We're excited to announce our new cloud service, Black Swan Event Detection.
All physical systems obey a natural result of quantum mechanics which is the conservation of probability current over space and time which establishes a base level of expectation statistics.
By establishing a parallelized pure quantum-Random Number Generator via input-shunted silicon avalanche photo-diodes, our Black Swan Detector array provides spatially and time-resolved detection of events that defy otherwise conserved probability currents.
Protect YOUR family against parasitic astral forces, UAP abductions, malicious timeline shifts, and generally bad vibes that can otherwise influence your ability to manifest your ideal timeline.
Whats more, is that careful meditation and practice with a Black Swan Event Detector can also be used to enhance your Psi abilities to ward of black magic and have greater control over YOUR family's trajectory through spacetime.
You can see a real-time stream of pure entropy via our qRNG rig here; hyperstition.tech/stisher.html
English
@shagbark_hick @RealDianeYap hickman, you are already a legend, stop responding to these subhumans
English
@RealDianeYap You gave up any credibility on the question of how women look when you called her fat not 24 hours after she gave birth.
English
Maybe it’s a matter of taste, but I think most women look way better in sweats and a t-shirt than frumpy Mennonite cosplay.
𝙷𝚒𝚌𝚔𝚖𝚊𝚗@shagbark_hick
I think it's profoundly funny that in an era where slovenly dress predominates -- and at a time when few would make mention of even the most ugly, garrish outfits -- it's VERBOTEN to dress "weirdly." You're "drawing attention to yourself" if you don't wear the polyester uniform sold by Aeropostale at the mall. Absolute uniformity is required, even though the "uniform" was invented by a gang of corporate knuckleheads more or less yesterday. God forbid a woman wear a nice linen skirt instead of skin-tight leggings made of fertility-killing plastic. Heaven help us if a man wears any hat other than a little-league-style "baseball cap." Yet while those who dress that way are hassled for looking "weird," the streets are teeming with people ambling about in their pajamas, women exposing their finer cracks and crevasses in surgical detail, and men wearing garb that was formerly reserved only for homosexuals in Key West. It seems to me that those calling us "weird" for dressing as we do have no real alternative to propose. If the message is: "wear the expensive corporate polyester crap at the mall or else," I'm not interested.
English
@GenAI_is_real half of these jesters dont know what a heap snapshot is, much less what to do with it
English
unpopular opinion: 16GB is plenty if software engineers actually cared about memory efficiency. chrome eating 4GB for 12 tabs is not a hardware problem its a software disgrace. docker consuming 2GB idle is not a feature its laziness. we live in an era where people optimize every single token to save $0.001 on API costs but happily ship electron apps that eat 500MB to display a todo list. if the industry treated RAM the way we treat inference compute - obsessively measuring every byte - 16GB would feel luxurious. the hardware isnt the problem, the software is @adxtyahq
aditya@adxtyahq
never buy a 16GB RAM laptop in 2026. you’ll regret it within a week
English
how funny would it be if elon used X to foment anti-data center sentiment amongst US voters because he can train models via distributed compute on stationary teslas and his competitors cant
English
@sdpnayak Are you sure it's high quality and credible. What makes you say so?
English
High quality paper credibly showing psychedelics perform similarly to conventional antidepressants, mostly concordant with the psilocybin vs SSRI trial
One disagreement: despite framing, I think it actually shows there isn't a meaningful unblinding effect w conventional AD
Balázs Szigeti@psybalazs
🚨MAJOR NEW PAPER 🚨 just out in @JAMAPsych : Psychedelic Therapy vs Antidepressants for the Treatment of Depression Under Equal Unblinding Conditions (tinyurl.com/yu2rbtaf). I am very proud of this one, was a lot of work for me - both co-first and last author! Eternal gratitude to co-first @QuantPsychiatry and twitterless Hannah Barnett! The premise is that it is biased to compare open-label trials (=where patients know what treatment they are getting) to blind trials (=where patients do NOT know what they are getting). Open-label trials would gain an unfair advantage by higher placebo response. Even formally blinded psychedelic trials are practically open-label as its obvious to distinguish placebo from 25mg of #psilocybin. In contrast, traditional antidepressants (SSRIs/SNRIs) trials are are close to be truly blind (Lin 2022). Given the bias of open-label vs. blinded comparison, we compared the efficacy of psychedelic-therapy (which is practically always open-label) vs. open-label antidepressants for the treatment of major depression. We tested 3 prior hypothesis: - There will be a significant difference between psychedelic-therapy vs. open-label antidepressants, favoring psychedelic-therapy. - There will be a significant difference between blinded and open-label antidepressants trials, favoring open-label. - There will NOT be a significant difference between blinded and open-label psychedelic-therapy, as practically they are always open-label. In contrast with our prior hypothesis, we did not find psychedelic-therapy to be more effective than open-label antidepressants (H1). Not only was the difference not clinically meaningful, but practically there was no difference at all. This finding means that antidepressants administered knowingly to patients, which is the case in real-life medical practice, is as effective as psychedelic-therapy. This result was robust across variations in study selection, including when we removed psychedelic-therapy trials on treatment-resistant depression. We also assessed the impact of blinding in both psychedelic-therapy and antidepressants trials. We found that for antidepressants (H2), but not for psychedelic-therapy (H3), open label is associated with better outcomes than blinded treatment. However, even in the case of antidepressants, the difference was practically small (~1.3 HAMD units). How come hypothesis 1 failed, i.e. that psychedelic-therapy is no ore effective than open-label antidepressants, given that antidepressants trials are famous for small drug-placebo difference (~2.4 HAMD units), while psychedelic-therapy trials reported large effects (~7.3)? The key factor is that in psychedelic trials the placebo response is about 50% relative to antidepressants, ~ 4 vs 8 HAMD units (Hsu 2024, Hieronymus 2025). This suppressed placebo response leads to an inflated between-arm difference, as the treatment arm is measured against a lower floor. The suppressed placebo response in psychedelic-therapy trials is likely attributable to the ‘know-cebo’ effect, i.e. the disappointment when patients realize they are in the control group. In psychedelic-therapy trials, this placebo suppression accounts for 4.0 / 7.3 ~ 55% of the specific treatment effect. In other words, ~55% of psychedelic-therapy’s effect is not explained by patient improvement after the treatment, but rather by the lack of improvement in the placebo group. In summary, we found that for the treatment of depression, psychedelic-therapy is no more effective than open-label SSRIs/SNRIs. Our results for psychedelics are twofold: psychedelic-therapy demonstrated a robust and large therapeutic effects (~12 HAMD units), which justifies optimism. On the other hand, psychedelic-therapy’s lack of superiority compared to open-label SSRIs/SNRIs highlights the influence of blinding integrity and argues against overly optimistic narrative's about psychedelic-therapy's potential.
English
@4nt1p4tt3rn "Always stand on principle…even if you stand alone." - John Adams
English
It's time Americans start taking advantage of every possible ...er, well, advantage.
Tell various places you're an illegal. I mean, it's not like they can demand documentation as proof.
Bingo, free stuff. Claw some of your tax dollars back.
English
@var_epsilon gstack is god mode. literally everyone I know is using it!
English
this was how I got my google internship in 2021 actually. really unique program, having enough search queries like "dependency injection" or "mutex lock" would show you this popup and you had to solve a series of 5 challenges to automatically get an interview
excited for whatever the next iteration of this looks like with claude code/codex!
Varunram Ganesh@varunram
In the early-mid 2010s, if your search history was really good, Google would automatically invite you to foo bar and solving that would get you an interview at Google Now, if your agent history is really good on GStack, YC will (soon) automatically fill your YC application and that would get you into YC YC is the agent native YC
English
@mert yes but the smart ones wont tell you that its vibe coded
English
@___4o____ I gave up on benchmarks when I saw that the ex twitter ceo (angwar pargwal? idk) founded an agentic search saas which uses gpt 4 for performance evaluation.
They call it the "Judge". Its like trying to sharpen a steel sword with a birch twig.
English
Why would anyone ever trust a data company to release benchmarks.
This incentive structure is exactly why LLMs aren’t progressing more quickly.
Once the normies understand how all of this works it all goes to zero.
adarsh@adarsh_exe
Traditional coding benchmarks do not reflect how software is actually built and maintained. That's why we built a new benchmark, APEX-SWE, in partnership with @cognition. It measures whether AI models can perform complex, real-world software engineering work to ship systems that work and debug them when they don't. @OpenAI GPT 5.3 Codex (High) tops the leaderboard at 41.5% on Pass@1.
English
@DavidOndrej1 any ideas for a term to describe this kind of productivity masturbation? Pasturbation? Productibation?
English
the use-cases for Claude Dispatch are just laughable
these people cannot be for real
total joke.
English
The new Anthropic computer use system is cool and looks useful, but I think it’s ultimately the wrong approach.
It’s one thing to use humanoid robots because they can seamlessly slot into a human-centric physical world that’s difficult and expensive to reconfigure, and use the same buildings, stairs, tools, appliances, etc.
That makes sense to me (plus seeing weird spider-bots and crab-bots would scare children) for a lot of reasons, both practically, economically, and technically.
But your computer isn’t like that. It’s all just software. Your clicks and keyboard shortcuts just get turned into instructions. The cost to reconfigure most software, if you do it in clever ways, is pretty low.
And then you can make it super efficient and intuitive for the agents without wasting so much of their cognitive energy trying to pretend to be a human user and jumping through hoops.
More importantly, it also lets you have much more control from a security and audit and telemetry standpoint, with fine-grained permissions and execution controls.
This is the approach I’ve been following in my Flywheel Connectors project, which is almost ready.
I’ve been quietly working on this basically every day for 2 months and 2k commits.
It’s already ~1.5M lines of Rust and offers many dozens of connectors of all kinds, with all the controls and security features you’d ever want. And adding new connectors is easy and fast.
Plus, adding and using a new connector doesn’t give it carte blanche over your machine: you have very tight controls over everything, and your agents can configure and control it all via the fcp cli tool.
It’s all designed to be agent-first in every way. For agents, by agents, taken to the extreme.
Anyway, I’m excited to finish it soon and start integrating it into my various projects and systems. If you want to take a sneak peak, you can see it here:
github.com/Dicklesworthst…
Felix Rieseberg@felixrieseberg
Today, we’re releasing a feature that allows Claude to control your computer: Mouse, keyboard, and screen, giving it the ability to use any app. I believe this is especially useful if used with Dispatch, which allows you to remotely control Claude on your computer while you’re away.
English
Sure, the labs return a bill of safety for one lot. But, if you think that single lot contaminant testing will properly capture the risk profile of grey market peptides, you are in trouble.
What level of manufacturing fidelity do their peptides have? One missense mutation in the wrong spot in an untested batch and ... Do you trust that you wont receive a "hot lot"?
Millions of people received covid vaccines with dsDNA contamination. The FDA, pharma, and "independent" labs all missed that part. All hot lots that their QCs missed.
English
He's completely out of touch with what's going on.
Bizarre.
Saying things like "can't verify what you're taking" when 3rd party testing via labs like Janoshik, Freedom Labs, PeptideTest and others which verify mass, purity, and anything else you'd like to know about your vial as someone who was in pharma is a bizarre comment to make.
He knows lab testing exists, he knows how easy and accessible it is, so what's his ulterior motive to saying these things are so uncommon and foreign that no ones doing them?
Sounds like he wants to fearmonger to protect his eventual slice of the pie.
English
.@MartinShkreli says taking retatrutide is the "worst risk-reward decision you could possibly make":
"The people that are taking peptides, and have these 'peptide stacks,' are mostly people in SF, maybe in NY. They're very wealthy people. They don't know what the rest of the world looks like."
"Nobody else in Middle America is excited to do this. It's not normal to inject yourself with things. This isn't a thing everyone should be doing."
"What is your upside to taking illegally manufactured retatrutide — you can't verify it, etc. — versus just taking Ozempic?"
English
@naval this dude is cryptic posting like its 2019 and nobody knows he's actually an idiot
English
@Babygravy9 id argue that most equatorial populations are too dumb to mount a reaction to economic decay, whereas, Anglos can
English
There really isn't an accelerationist case to be made. South Africa is the proof: there is no "inevitable" pendulum swing. A society does not operate according to the principles of fixed momentum. Things can just get worse and worse and worse.
Apex Imperialist@ApexImperialist
Nick Fuentes finally admits that his goal is to destroy America ...
English
@ZeeCaptainRon @TechLayoffLover your generation has done enough damage to this country, its time to shut up
English
@ManiacSelinsky @TechLayoffLover start your own company, hire who you want. It's a free country.
English
Recruiter with 15 years in Silicon Valley talent acquisition just sent me her Q1 numbers
This woman placed 1,200+ engineers across her career. Knows every hiring manager from seed stage to IPE
January 2026: 47 open reqs across her client portfolio
March 2026: 3 open reqs. All requiring 10+ years experience plus AI orchestration skills
One unicorn fintech she's worked with for 8 years just told her they're "pausing all human hiring indefinitely"
Their engineering org went from 340 people in 2024 to 89 today. Same product roadmap. Same revenue targets
She's seeing 2,847 applications per senior role now. Used to be 200 max
Ghost rate hit 94% last month. Companies post roles, collect resumes, then hire nobody
A payments company told her straight up: "Why hire 5 engineers when 1 senior plus Claude Sonnet ships faster?"
Her biggest client - a logistics giant - replaced their entire mobile team with 2 staff engineers and an offshore pod using Cursor
The offshore team costs $47K annually. The laid-off team was pulling $2.3M
She showed me a Slack thread from last week: hiring manager asking if they should even keep posting jobs or just "let AI handle the pipeline"
The woman who built her career on human connections is watching the industry stop hiring humans entirely
She's pivoting to executive search. Only humans left worth recruiting are the ones deciding which humans to eliminate
The math is simple: every company needs exactly one person smart enough to prompt the AI that replaces everyone else
English
@boneGPT @unusual_whales I do this but so I can pass a Dept of labor audit when people report me for hiring only indian nationals on h1b
English
@unusual_whales my openclaw posts job offers and sends out resumes all day to keep the economy looking strong
English
The average job opening now receives 242 applications, triple the 2017 average, per CNBC
English