Competing for Soft Power by Maria Repnikova
Repnikova examines China's ambitious but uneven image-making in Africa, and the role of the West in mediating this encounter.
https://t.co/KA5ItmIVbM
Mostly some that have been fundamental in the development of my political thought after being for years in the intellectual swamp of classic Stalinism/Maoism.
The 6 Processors Powering Modern AI
CPU — Orchestration & Preprocessing (Best for general tasks and OS management)
GPU — Training & Deep Learning (Best for massive parallelism and CUDA workloads)
TPU — Google-Scale Tensor Workloads (Highly efficient for large neural network scales)
NPU — Edge & Mobile Inference (Low power consumption directly on mobile hardware)
LPU — Real-Time LLM Serving (Ultra-low latency inference using on-chip SRAM)
DPU — Data Center Infrastructure (Frees up CPU by handling network traffic and storage)
Navigating hardware efficiency means selecting the specialized processor designed exactly for your workload.
El Estado moderno
Basado en la teoría del Estado continental europeo, este libro de Vesting sostiene que el Estado occidental moderno tiene tres capas superpuestas: el Estado constitucional, el Estado de bienestar y el Estado en red
Información y Cap 1: https://t.co/JW1VPIq2E2
10. The part that stings
Every one of these defaults exists because it benefits Netflix's engagement numbers, not your experience. Autoplay increases watch time. "Auto" quality saves them bandwidth. A messy Continue Watching row means more potential clicks. Stale viewing history still drives recommendations even when it's wrong.
Her last line: "None of this is broken. It's working exactly as designed. It's just designed for their metrics, not your Tuesday night."
7. She cleared his Continue Watching graveyard
18 titles sitting in that row — shows abandoned three minutes in, a documentary he misclicked, half a season he forgot about. All of it pushing real recommendations further down the page.
Three-dot menu on each title → "Remove from Continue Watching." Two minutes, gone.
Netflix doesn't auto-clean this row because more items sitting there is more potential taps — clutter serves their engagement numbers, not your browsing experience
5. She showed him the hidden genre codes
She typed https://t.co/Lc8ZRIYZcT… into his browser. A wall of thrillers he'd never seen on his homepage appeared.
Netflix organizes its full catalog with over 2,000 hidden category codes — the homepage only ever shows the ~30 rows the algorithm thinks will keep you scrolling longest. The rest exists, fully browsable, if you know the URL trick.
She bookmarked the genre index page for him. Told him to search it whenever the homepage felt empty.
4. She reset his entire recommendation profile
Account → Profile → Viewing Activity → Hide All.
Everything he'd ever watched — a rom-com his ex picked, three seasons of something he fell asleep to constantly, his nephew's cartoons from a holiday visit — was still shaping his 2026 recommendations.
One button wiped it. The algorithm rebuilds itself over the next 48 hours based only on what he watches going forward. Free reset, no new account required.