Raphael Spannocchi

15.6K posts

Raphael Spannocchi banner
Raphael Spannocchi

Raphael Spannocchi

@raphbaph

How can democratic decisions be smarter than the median voter? Beyond Arrow's paradox. Building @Anode_GG.

Text Editor, mostly Katılım Ocak 2010
846 Takip Edilen1.8K Takipçiler
Sabitlenmiş Tweet
Raphael Spannocchi
Raphael Spannocchi@raphbaph·
The true test of any organizational paradigm or collective consensus building effort is this: Does it make the whole more intelligent than the one, or less? Do people get better when interacting with the system? Or worse? It's really that simple, and that hard.
English
5
1
31
10K
Colin Wright
Colin Wright@SwipeWright·
It's like if Henry Ford made his first cars mimic the feel of horses and smell like manure.
Sawyer Merritt@SawyerMerritt

Lexus says its upcoming all-electric LFA supercar will use vibrations and other sensory feedback to mimic the feel of a combustion engine. "We’re not just wanting to replicate the sound of the engine: we want to redesign the sound itself so the car feels more alive and authentically connected." The production version of the car is expected to be unveiled at the end of this year/early next year. Concept unveiled in Dec 2025:

English
31
22
513
19.5K
Raphael Spannocchi
Raphael Spannocchi@raphbaph·
@Austen Actually think they're still profitable on Max even. Waiting for Grok and Meta to catch up and drive prices to 1/10th. Then we will see true subsidies
English
0
0
2
68
Austen Allred
Austen Allred@Austen·
Yes. When talking about subsidization of max plans there are three variables to consider: 1. What the inference cost is to the AI company 2. What people are willing to pay for tokens on the margin 3. How well AI companies will be able to price discriminate with regard to 2
Blake Anderson@blakeandersonw

The costs of “llm usage subsidization” are completely overblown. If a $200 max plan equates to $4000 in credits, the mean subscriber may spend $1500. Cost of inference is probably around $300. The additional $100 purchases data, mindshare, and likelihood of api usage.

English
9
2
26
15K
Raphael Spannocchi
Raphael Spannocchi@raphbaph·
Worked as a sound engineer for a decade. this is unlikely to succeed. let's look at the physics: for sound cancellation to work the same waveform has to reach the ear of the listener with the same amplitude but reversed polarity. This is somewhat doable with headphones because their position is fixed. For higher frequencies, say 2kHz, where a lot of the painful noise is, wavelength is about 17cm (8in) So your speakers woukd have to reach the ears of 100s, or even 1000s of listeners with incredible accuracy. Just very tough. Easier would be to mount these devices on the emitters. Buses, trucks, etc. Has been done on choppers with some success
English
0
0
0
22
Jay Hack
Jay Hack@mathemagic1an·
Big assumption I’m making here is the sound you want to cancel is simple and predictable, eg a chainsaw or a bus You can infer this audio from video with current foundation models. Also not a huge issue to allow temporary spikes in volume But agreed it is a challenging aspect
English
9
0
26
4.8K
Jay Hack
Jay Hack@mathemagic1an·
What if we could cancel out urban noise pollution? Place a set of speakers that act like noise-cancelling airpods for all pedestrians simultaneously You can actually do this - and it paves the way for an "acoustics foundation model" Simulation code + proposal 👇
English
82
39
893
429.8K
Ben Tossell
Ben Tossell@bentossell·
Sol sunset?! wtf
Ben Tossell tweet media
English
13
0
79
25.8K
Raphael Spannocchi retweetledi
maria
maria@maria_rcks·
I solved the model picker problem, no need to thank me Tibo
English
392
340
10.3K
1.4M
Leuts.eth
Leuts.eth@A_Leutenegger·
Apparently airdrop farmers are still around!
English
4
0
3
621
Florian Glatz
Florian Glatz@heckerhut·
@raphbaph good questions. the answer to all of them is of course freedom browser :-) we want to make this super easy for anyone. still not there yet - but in about 6 months I’m sure we‘ll be there, including fiat onramping.
English
1
0
7
62
Florian Glatz
Florian Glatz@heckerhut·
publishing on the dweb is genuinely superior to publishing on web2: 1) purchase a stamp on swarm 2) publish content 3) point a .gwei domain at the hash 4) profit the experience is so fast & frictionless, it’s mind-boggling how good it is how have we not won yet?
English
3
2
27
1.4K
Auryn
Auryn@auryn_macmillan·
🤔 @theInterfold needs stablecoins in two places. 1. Fee token that used to request a job 2. stable portion of the node bond They do not need to be the same token. USDC is out, along with any other tokens that can arbitrarily freeze accounts. Which tokens should we choose?
English
16
4
36
2.3K
Raphael Spannocchi
Raphael Spannocchi@raphbaph·
Okay: From the study, source Cambridge Out of 27 BH members only 15 said they actually used AI Of those only 5 could identify common AI Logos such as Anthropic and OpenAI! They said they used DeepSeek in 23-24, when DS came out in 24 "Bomb building" and "attack planing" are taken at face value and can in fact not be corroborated. This is a breathless hit piece.
English
0
1
7
1.1K
Antonia Juelich
Antonia Juelich@AntoniaJuelich·
In a hotel room in northeast Nigeria, I opened a leading AI chatbot, turned my laptop toward a former Boko Haram commander, and asked if he'd used it. He nodded. "You type in the question… like 'How can I build a bomb?', and then it tells you how. It is like a human robot. We used it a lot." My new study on how the jihadist terrorist group Boko Haram uses frontier AI with @CamAISciPolicy, covered today in @nytimes 🧵/9
English
281
1.8K
7.6K
2.5M
emozilla
emozilla@theemozilla·
We're working on making the local model experience better in Hermes, what are the best local models at each weight class? My blind guess, please correct: 8-16 GB VRAM Gemma4 12B 24-32 GB VRAM Qwen3.6 27B Qwen3.6 35B 128 GB VRAM (Spark, M3 Max) ??? Can you do DSv4-Flash?
English
125
13
258
63.5K
𝙾𝚕𝚞𝚝𝚊𝚢𝚘
𝙾𝚕𝚞𝚝𝚊𝚢𝚘@OlutayoBamz·
@raphbaph @AntoniaJuelich Well, it's believable. A lot of coverage have gone into similar research on terrorism in Nigeria. According to her, she "sat" with former Boko Haram members who have most likely denounced allegiance to the group (forever skeptical about Nigeria's rehab ethos). 1/3
English
3
0
3
147
Raphael Spannocchi
Raphael Spannocchi@raphbaph·
@GasiorekM Think it's mostly fabricated. Obviously they use AI. But this smells "we tell it to build a bomb for a movie " worked in 2023 maybe
English
1
0
0
73
Michael G
Michael G@GasiorekM·
What we biddin' up on this AI adoption news, boys?
Michael G tweet media
English
1
0
0
94

Keşfet

@SwipeWright @grumpyinnovator @Austen @bentossell @A_Leutenegger @heckerhut @auryn_macmillan @theInterfold