Who this is
One person, a short shelf of apps
Pea Kirste is the name on the software; Per Andreas Kirste is the person behind it, working from Sarbuvollveien 9B in Høvik, just outside Oslo.
There is no team, no roadmap on a wall and no growth target. An app gets made because it is worth making, then it gets looked after.
What that buys you
A one-person shop cannot run a data operation even if it wanted to, and this one doesn't want to. No accounts to manage means no database to leak, no login to forget and no service that stops working the day the bills go unpaid.
How the apps work
Built to be finished, not fed
Everything an app needs to remember goes into its own private folder on the phone, walled off by Android from every other app installed. No account is created, no identifier is minted, no event is counted anywhere, and no advertising library is compiled into the build.
The upshot is an app that still works years from now on a phone with no signal, and a privacy policy that is mostly a list of things which never happen. Both of those are easier to maintain than the alternative.
Published by Pea Kirste
Tanglewire
com.knotgrid.tanglewire
A puzzle about pulling order out of a mess. Each level throws down a web of glowing nodes joined by wires, crossing each other everywhere, and the job is to drag the nodes about until not one wire overlaps another.
Crossings are counted live as you move, wires glow hot where they still cross and cool when they come clear, and when the last one slips free the whole web settles and lights up. Dozens of hand-built levels run from a gentle handful of nodes to dense webs, with a hint that nudges a stuck node, a reset that re-scrambles the same graph, and a free-play mode that spins up a tangle of whatever size you fancy. It runs offline, in English and Norwegian, with no account and no advertising.
Tanglewire on Google PlayWhy a thrown-down web looks so much worse than it is
Two wires that share no node cross with a probability that has been known exactly since 1864, when Sylvester asked how often four random points sit in convex position. In a square the answer is 25/36; of the three ways to pair four such points into two segments, exactly one pairing gives the diagonals, so any two disjoint wires cross with probability 25/108 — near enough 23%. Multiply that by the number of disjoint pairs and you have the size of the mess before you touch anything. The other half of the panel is Euler's V − E + F = 2, which caps a crossing-free drawing at 3V − 6 wires however cleverly it is arranged.
188pairs of wires sharing no node
43.5crossings expected from a random throw
23.1%chance any one such pair crosses
39wires a flat drawing can ever hold
105wires if every node joined every other
37%of those could be drawn without crossings
Not on this panel: no board, no nodes to drag, no graph drawn and nothing untangled. It counts what a random scatter costs you and where the ceiling is; which node to move is not a question arithmetic can answer.
The promise under the game
Every level really can be untangled
A level is built as a planar graph — one that can be drawn flat without crossings. That it can always be done with straight wires, rather than curved ones, is a theorem: Fáry proved it in 1948, and Wagner and Stein had it too.
So a board is never a trick. Whatever state it is in, some arrangement of those same nodes has no crossings at all, and the only thing standing between you and it is where everything happens to be sitting.