<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Field Notes</title>
  <subtitle>A working journal on inference, infrastructure, and the quiet mechanics of decision-making at scale.</subtitle>
  <link href="https://khramov.io/atom.xml" rel="self"/>
  <link href="https://khramov.io/"/>
  <id>https://khramov.io/</id>
  <updated>2026-04-19T00:00:00.000Z</updated>
  <author><name>Oleksii Khramov</name></author>
  <language>en-gb</language>
  <entry>
    <title>Leaflet &amp; openstreet map</title>
    <link href="https://khramov.io/writing/leaflet-test/"/>
    <id>https://khramov.io/writing/leaflet-test/</id>
    <updated>2026-04-19T00:00:00.000Z</updated>
    <summary>Exploring leaflet maps</summary>
    <category term="maps"/>
  </entry>
  <entry>
    <title>On legibility, and the cost of making things clear.</title>
    <link href="https://khramov.io/writing/legibility/"/>
    <id>https://khramov.io/writing/legibility/</id>
    <updated>2026-04-14T00:00:00.000Z</updated>
    <summary>Every interface that explains itself is also an interface that decides what can be explained. The cost is rarely paid by its author — and that is the part worth writing down.</summary>
    <category term="essay"/>
    <category term="systems"/>
  </entry>
  <entry>
    <title>Infrastructure as assumption</title>
    <link href="https://khramov.io/writing/infrastructure/"/>
    <id>https://khramov.io/writing/infrastructure/</id>
    <updated>2026-04-10T00:00:00.000Z</updated>
    <summary>The infrastructure we build is never just infrastructure. It is frozen policy. It is calcified decisions, made in specific contexts by specific people, now invisible to everyone who uses it.</summary>
    <category term="systems"/>
    <category term="infrastructure"/>
  </entry>
  <entry>
    <title>Quiet defaults.</title>
    <link href="https://khramov.io/writing/quiet-defaults/"/>
    <id>https://khramov.io/writing/quiet-defaults/</id>
    <updated>2026-04-02T00:00:00.000Z</updated>
    <summary>A defense of software that does nothing until asked. Notes from three years of stripping features out of the same tool.</summary>
    <category term="notes"/>
    <category term="systems"/>
  </entry>
  <entry>
    <title>The observer is never the observed.</title>
    <link href="https://khramov.io/writing/observers/"/>
    <id>https://khramov.io/writing/observers/</id>
    <updated>2026-03-20T00:00:00.000Z</updated>
    <summary>Why dashboards lie, and what kind of lying is actually useful. A short taxonomy of metrics that change the thing they measure.</summary>
    <category term="inference"/>
    <category term="essay"/>
  </entry>
  <entry>
    <title>Field notes VII — a week without inbox.</title>
    <link href="https://khramov.io/writing/field-notes-vii/"/>
    <id>https://khramov.io/writing/field-notes-vii/</id>
    <updated>2026-03-11T00:00:00.000Z</updated>
    <summary>Seven days of treating every message as optional. Some observations on attention, expectation, and the social contract of replies.</summary>
    <category term="field"/>
    <category term="notes"/>
  </entry>
  <entry>
    <title>Compression as a moral act.</title>
    <link href="https://khramov.io/writing/compression/"/>
    <id>https://khramov.io/writing/compression/</id>
    <updated>2026-02-28T00:00:00.000Z</updated>
    <summary>What you leave out is an argument about what matters. A piece on editorial restraint, from newsrooms to model weights.</summary>
    <category term="essay"/>
    <category term="inference"/>
  </entry>
  <entry>
    <title>Second-order effects and the people who budget for them.</title>
    <link href="https://khramov.io/writing/second-order/"/>
    <id>https://khramov.io/writing/second-order/</id>
    <updated>2026-02-17T00:00:00.000Z</updated>
    <summary>A short essay on planning horizons — why most organizations plan for the first-order effect and contract out the rest.</summary>
    <category term="systems"/>
    <category term="essay"/>
  </entry>
  <entry>
    <title>Seams.</title>
    <link href="https://khramov.io/writing/seams/"/>
    <id>https://khramov.io/writing/seams/</id>
    <updated>2026-02-05T00:00:00.000Z</updated>
    <summary>On the lines where systems meet — where responsibility is usually dropped. A case study from a logistics platform.</summary>
    <category term="field"/>
    <category term="systems"/>
  </entry>
  <entry>
    <title>In praise of slow inference.</title>
    <link href="https://khramov.io/writing/slow-inference/"/>
    <id>https://khramov.io/writing/slow-inference/</id>
    <updated>2026-01-24T00:00:00.000Z</updated>
    <summary>Real-time answers are easy to sell and hard to justify. What changes when a model is allowed to take its time.</summary>
    <category term="inference"/>
  </entry>
  <entry>
    <title>What we counted, and what the data said.</title>
    <link href="https://khramov.io/writing/what-we-counted-data/"/>
    <id>https://khramov.io/writing/what-we-counted-data/</id>
    <updated>2026-01-19T00:00:00.000Z</updated>
    <summary>A look at how resolution rates, recurrence patterns, and ticket volume interact — and what a year of data actually showed.</summary>
    <category term="field"/>
    <category term="notes"/>
    <category term="systems"/>
  </entry>
  <entry>
    <title>What we counted, and what we meant to.</title>
    <link href="https://khramov.io/writing/what-we-counted/"/>
    <id>https://khramov.io/writing/what-we-counted/</id>
    <updated>2026-01-12T00:00:00.000Z</updated>
    <summary>A postmortem on a metric that accidentally ran a team for eighteen months.</summary>
    <category term="field"/>
    <category term="notes"/>
  </entry>
  <entry>
    <title>Notes on working memory.</title>
    <link href="https://khramov.io/writing/working-memory/"/>
    <id>https://khramov.io/writing/working-memory/</id>
    <updated>2025-12-29T00:00:00.000Z</updated>
    <summary>Tools I keep on my desk. Tools I've learned to put away. Some opinions on ambient context.</summary>
    <category term="notes"/>
  </entry>
  <entry>
    <title>Permissioned rooms.</title>
    <link href="https://khramov.io/writing/permissioned-rooms/"/>
    <id>https://khramov.io/writing/permissioned-rooms/</id>
    <updated>2025-12-18T00:00:00.000Z</updated>
    <summary>A look at access-control patterns in physical and digital spaces, and what they reveal about who is trusted with what.</summary>
    <category term="systems"/>
    <category term="essay"/>
  </entry>
  <entry>
    <title>Residue.</title>
    <link href="https://khramov.io/writing/residue/"/>
    <id>https://khramov.io/writing/residue/</id>
    <updated>2025-12-02T00:00:00.000Z</updated>
    <summary>Every process leaves traces. A short piece on audit logs, telemetry, and the quiet accumulation of record.</summary>
    <category term="inference"/>
    <category term="systems"/>
  </entry>
</feed>