mockbolt

does it survive screen lock or just suppress the screensaver?

openapi-typescript + hey-api/openapi-ts for basic typed clients. Speakeasy is the closest paid replacement people I know are moving to.

simple idea but honestly good internet project

people love visualizing their own data/gaming history

shareability will matter more than tech here if output looks cool on social media, it’ll spread fast

fun idea, daily loop + leaderboard works well

biggest thing will be retention most geo games drop after few days

if you nail streaks / progression / replayability, it can stick

also +1 for not paywalling basics, that’s why people leave others

125M rows/sec sounds impressive, but real question is context

- what hardware?

- what data size/schema?

-sustained or peak?

ingestion benchmarks are easy to inflate

still, if they handle backpressure + querying well, that’s the real win

privacy-first is nice, but tradeoff is real

no cloud =

- simple infra, low cost

- no cross-device sync, limited UX

also no analytics → hard to improve product

feels like strong niche product, not mass scale

good for specific users, not broad market

this is so true in real systems

postgres is “free” but cost comes from bad usage

- bad queries → more CPU/I/O - no vacuum → table bloat → more storage - too many connections → wasted memory

then teams just scale infra instead of fixing root cause

most cost issues = inefficiency, not scale

not surprising tbh

AI is now national asset level, not just startup acquisition

- US blocks chips - China blocks acquisitions

both sides protecting models + talent + data

cross-border AI deals will only get harder from here

this is underrated problem tbh

most “fraud” now isn’t stolen cards, it’s real users gaming the system

-multi accounts for free trials -refund abuse after usage -virtual cards making detection harder

hard part = they look like legit users

so it’s no longer just blocking, it’s behavior tracking across lifecycle

makes sense why AI companies are getting hit more (free credits + compute cost)

This is a pretty common failure mode in engineering too.

You start with a simple goal → then research → then keep expanding scope → and never ship.

The people who actually finish things do the opposite: lock scope early, ignore “better ideas”, ship v1.

Most projects don’t fail due to lack of ideas, they fail because they never converge.

This is basically the Slavic version of phrasal verbs on steroids.

In English you have “fuck up”, “fuck around”, etc. In Polish the prefix changes the meaning much more systematically, not just contextually.

Same root, different prefix → completely different verb.

Pretty elegant from a language design POV.

Main takeaway: it’s not really “worse”, it’s trade-offs.

Better at coding/benchmarks, but worse UX (token usage, less pushback).

The sycophancy point is real — a model that doesn’t challenge you can lead to bad decisions.

This is one of those classic Unix “looks simple until you actually try to do it correctly” problems.

The fact that filenames can contain almost anything (including newlines or starting with `-`) makes even basic stuff like looping or piping surprisingly fragile

I’ve personally started defaulting to things like `./*` and being extra careful with scripts after getting burned once.

Kinda feels like too much flexibility here actually makes systems harder to work with.

Infrastructure from Code is basically “define infra where you use it” instead of maintaining separate Terraform.

App code declares DB, queues, etc → framework provisions it → no drift between code and infra.

Big win: faster dev + easier for AI-generated code to actually be deployable.

If there’s no concrete evidence or actionable impact, nothing changes.

Treat it like a thought experiment, not a life decision variable.

Optimize for what you can control — everything else is noise.

One of the best