the nostalgia for 1999 isnt really about missing dial up or basic html, its about mourning the loss of user agency. back then the browser actually worked for you. today the client is basically a hostile enviroment running megabytes of third party js just to track telemetry.
going back to gopher or text-only browsers feels like admitting defeat tbh. we can still build incredibly fast modern apps if we just stop treating the users hardware like an infinite resource for adtech. you dont need massive frameworks and client--side bloat to make something good.
charging an enterprise premium just to give users the privlege of opting out of metadata harvesting is wild. they know exactly how valuable project structures, issue titles and sprint cadence data are for training models.
this trend is exactly why relying on client-side 'trust" or saas toggles is failing us. if you want real privacy now you have to build it into the architecture. aggressive edge caching and stateless proxies that sanitize payloads before they even hit the upstream provider is basically mandatory now. if the data never reaches their persistence layer, they cant train on it. we need to start trusting our own infra, not their updated tos.
the npm supply chain attacks were a massive wakeup call. the fact that we normalized storing sensitive tokens in localstorage for the last decade is wild.
moving to a bff pattern isnt just about hiding tokens, its about reducing the client attack surface entirely. shifting api orchestration and sanitization to edge proxies makes so much more sense. the browser should just be a dumb terminal rendering ui, not a secure vault managing state and credentials
he silent downgrades are the worst part. dropping the cache ttl from an hour to 5 minutes and spinning it as an optimization is just insulting. relying on their official clients feels like a trap now since they will always throttle the ux to save on their own compute costs. building any serious daily workflow on top of these managed tools right now just feels like building on quicksand.
exactly. calling it 'anonymized' is pure security theater once you have enough data points to map out someones daily routine.
waiting for legislation or eulas to fix this is a lost cause since adtech always finds a loophole. the fix has to be architectural. moving toward stateless proxies that strip device identifiers at the edge before they even hit upstream servers. if the payload never touches a persistent db there is literally nothing to de-anonymize. stateless infra is the only sane way forward
w3m is still the goat for quick lookups when you dont want to open a browser and lose 2gb of ram instantly. funny how we need forks of text browsers in 2026 just to keep the web readable. really like the addition of gemini support here...
definitely giving this a spin on my dev box later. the fact that it still uses a quilt patch series for dev is such a throwback tbh.
id 100% read that post. the jump from free tier serverless to why is my aws bill $400 this month for a hobby project is a rite of passage at this point. a $5 hetzner or digitalocean box with dokku/docker-compose is basically a superpower that most newer devs just bypass entirely now.
i think you missed the "on day 1" part of my comment. k8s, iac, and observability are incrdible tools when you actually have the scale and team to justifiy them.
my point is strictly about premature optimizaton. ive seen teams spend their first month writing helm charts and terraform before they even have a single paying user. if you have product-market fit and need zero-downtime rollbacks, absolutly use k8s. but if youre just validatng an mvp, a vps and docker-compose (or sqlite) is usually enough to get off the ground.
props for actually publishing it tbh. transparent engineering takes are so rare now, usually its just seo fluff.
weve basically been brainwashed to think we need kubernetes and 3 different databases just to serve a few thousand users. gotta burn those startup cloud credits somehow i guess.
mad respect for the honesty though, actually makes me want to check out db pro when i finally outgrow my flat files.
people wildly underestimate the os page cache and modern nvme drives tbh. disk io today is basically ram speeds from 10 years ago. seeing startups spin up managed postgres + redis clusters + prisma on day 1 just to collect waitlist emails is peak feature vomit.
a jsonl file and a single go binary will literally outlive most startup runways.
also, the irony of a database gui company writing a post about how you dont actually need a database is pretty based.
vifm is great if you live in vim, but mc is just hardcoded into my brain at this point. totally agree on zoxide + fzf though, cant imagine navigating a large fs without them now. they're the only 'modern' additions to my workflow that actually felt worth the setup time.
this hits home. ai makes it way too easy to ship 'more' without asking if we actually should. we've been in feature vomit mode for years now, shipping mb's of js for the most basic crud tasks. simplicity is becoming an act of engineering discipline rather than just a design choice tbh. less is actually more
still haven't found anything that replaces mc for me. the 2-pane layout is basically muscle memory at this point. everything modern just feels way too bloated or slow. mc is great but customizing the bindings is a total headache tbh. really like the idea of better vim integration here. curious how it handles performance on large directories with 10k+ files? giving it a spin...
seeing that Panda-70M research paper linked above makes this even crazier. the 'good faith' part of DMCA is basically never enforced in reality. platforms have used the 'shoot first, ask questions later' approach for so long that individual creators are just collateral damage. its about time someone actually challenged this in court. the power imbalance here is just wild tbh.
Total convenience vs security facepalm. We audit Signal’s crypto but the OS just pipes the plaintext into a local notification DB by default. If data isnt persisted, it cant be forensically extracted later. Honestly, sometimes the best privacy feature is just not having a database at all.