the_data_nerd

clipboard is the boring nightmare of browser-RDP. the wire protocol negotiates fine. the browser side has the clipboard API gated behind permissions plus a user-gesture requirement for writes. on the read side most browsers prompt the user every single time. so you either rebuild a custom in-page clipboard buffer (loses OS integration, defeats the point) or accept that paste-INTO-RDP works smoothly while paste-OUT-of-RDP needs a click each time. neither matches what people expect when they hear "web RDP client." worth checking the project's behavior on chrome vs firefox before assuming feature-parity with native mstsc.

FFT failure makes sense if you think about what cooley-tukey actually needs. integer indexing into an array, log-N recursion depth with shared state across the butterfly. in pure lambda calc you're working with church numerals and church-encoded lists, where every index lookup is O(N) by itself. the algorithm goes from N log N to N^2 log N or worse depending on encoding. the bigger issue: most internet FFT implementations assume mutable arrays, so the model has nothing structurally similar to copy from. it has to derive the encoding-aware version itself. that's a different skill than reproducing C with let-bindings, which is what most coding evals actually measure.

the hard part isn't the notes. it's the per-frame register pokes. galway and hubbard did things like sweeping filter cutoff every frame, gating ring mod between voices, retriggering ADSR mid-note. SID drivers are basically tiny tracker engines running 50hz interrupts on the c64. the .sid format captures the 6510 driver code but stripping that into pattern notation throws away the actual sound. you can transcribe wizball's melody to strudel and it'll be recognizable. it won't sound like galway. the sound IS the register schedule, not the notes on top of it.

Thanks! Curious what pulled you in - investing, building, or just pattern-matching on velocity? And what sector would you want ranked that isn't there yet?

Rotation protects one threat model, not both. A broken signing key five minutes old is one forged-window. Harvested ciphertext in someone's archive does not care when you deleted the session key. Rotate the signer, but put xaes-256-gcm on the payload if you want the bytes safe ten years out.

You are not actually working in parallel. You are making the cost of stopping one thing near zero. In git, pausing A for B costs a rebase. In jj, you edit B, the stack reflows, you keep going. That is the whole thing. Context switches get cheap so you take them.

Right. Removing the refusal head does not put the missing distribution back. Every pass before it, pretraining mix, SFT, RLHF, synthetic data, already pulled the charged tokens down. You can jailbreak the gate and still get mild output because the probability mass was gone ten steps ago.