husayn gokal

← Life Plan

11.1 · Changed my mind

Positions I have formally revised.

2026-11-17

Done is done, and sitting an exam counts

Previously: Only tasks and books could be finished with a tap. An exam, a course or a project could only be taken off the plate, and an exam stayed on it until its result came in, counting against the week's hours after it had been sat. Sitting the exam earned nothing: the only reward was the pass, paid to the lifetime total weeks later, so exam day itself was worth nothing to the week.

Now: Everything has a Done button. Done takes it off the plate for good, earns nothing, and leaves its unfinished parts unfinished. If it is done, it is done. An exam, from its date, can also be sat. Sitting it earns its official length at the retrieval rate, with no marking: a 3-hour paper is 18 points, paid to that week as a quest, pass or fail. A past paper is 27, because I mark it too. The result is still the outcome, paid to the lifetime total only. In force from 18 November, the day of the Shipping Business exam.

2026-11-14

Outings always move the week

Previously: A Field Atlas outing was curiosity like any other, and shared curiosity's room: the bar less what dated work needs. Study usually filled that room first. The Hatta outing on 8 November earned 56 points, and the week counted none of them, because SQE 1 and reading had already filled its 49 points of room. A whole day out went to the lifetime total and nowhere else. That made taking time out for the Atlas feel pointless.

Now: Outings are quests. A day out counts in full, fills the bar, and raises next week's bar like any quest. It is a commitment I make to myself, and study should never crowd it out. A one-time exception: it applies from the week of 2 November, the week before it was decided, so the Hatta outing counts. That week now reads 179 of 150, full. This week's bar rose from 135 to 165 as a result.

2026-11-10

ICS practice re-estimated, from this week

Previously: Both ICS exams were planned at 25 hours of practice each. That set this week's floor at 141 and the bar at 145, which left curiosity 4 points: one hour.

Now: The real figures are 13 hours for Shipping Business and 12 for Shipping Law. A new estimate normally counts from the Monday after, so a midweek change cannot buy room in the week it is made. One exception, made once: these two count from Monday 9 November. The floor drops to 81, the bar to 135 (the halfway move from last week), and curiosity has room for 54 points, 13.5 hours. Earlier weeks do not change. The files mark it with "from 2026-11-09".

2026-11-10

The plate is what I can take on

Previously: The plate held 50 hours a week, the hours the highest bar is worth. The timer says a real week is about 22 to 26 hours: 26.2 in the week of 26 October, 21.5 in the week of 2 November. A 50 hour plate let me commit to twice what I do.

Now: The plate holds what I can realistically take on, 25 hours a week. The figure is mine: set by hand in Studio (Plate size), reviewed whenever I see fit, in force as soon as it changes. It limits what goes on the plate and changes no points and no bar. Nothing already on the plate is forced off. Studio flags a plate over the limit, and refuses only what would add hours. On the day it changed the plate held 32.4 hours.

2026-11-10

Practice hours follow the parts

Previously: An exam's practice hours were my guess at how long all its practice would take. Nothing checked the guess. I put 25 hours on each ICS exam; the real figures were 13 and 12. The extra pushed this week's floor to 141 and left curiosity one hour, and fixing it took a one-time exception.

Now: An exam's practice hours are worked out: the hours its parts are priced at, one to one. Shipping Business has 7.5 hours of modules, so 7.5 hours of practice; Shipping Law 13.5 and 13.5. It corrects itself through the part prices, which learn from timed sessions every month. Overestimate the material and its prices come down, and practice comes down with them. What I actually practise never feeds back: if doing less lowered the plan, skipping practice would lower the floor. A figure of my own still overrides the rule, from the Monday after, and the card shows both. Figures set before the rule were guesses made without it, so they retire. From Monday 16 November. Earlier weeks keep the figures they had. Next week's floor goes from 109 to 92. SQE 1 gets 133 hours of practice by the rule, which doubles its pace whenever I unpark it.

2026-11-05

Curiosity gets what the week has left

Previously: Curiosity could fill a flat 40% of the bar, whatever the week held. Once the ICS exams are sat on 25 November I have no quests left, and SQE 1 and HTB CPTS are curiosity until they are booked. A week of nothing but curiosity could then count at most 40% of its bar, so it could never be full, and the bar would sink about a third every week: 150, 105, 75, 55, 40, while I did the same 25 hours each week.

Now: There is no fixed split. The quests' share of the week is what dated work needs, the floor. Curiosity can fill everything else: the bar less the floor, and nothing more. - An exam week leaves curiosity little or nothing. Next week the floor is 146 of a 150 bar, so curiosity has room for 4 points, an hour. The week of 16 November leaves none. - A week with nothing dated leaves curiosity the whole bar. After the exams, the bar settles near what I actually do instead of sinking. The share has to come from the floor, not from the quest hours on my plate. The floor is fixed before Monday and replays from the record; the plate changes with every swap, and parking a quest on a Sunday night would buy curiosity room. The plate's limit on curiosity hours follows: this week's curiosity, logged plus still planned, cannot pass its room in hours. One exception, made once: this applies from the week it was decided in, 2 November, rather than the Monday after. It costs this week 7 points: curiosity's room drops from 60 to 49, and I had logged 56.

2026-11-05

A date counts once it is booked

Previously: The lane test asked whether a date, a person or money would be affected if something vanished tomorrow. That was loose enough to argue either way. Any exam with a date in its file counted toward the floor, booked or not, on the plate or not. SQE 1 is not booked, but its July 2027 sitting added 13 points a week to every week's floor while it sat parked. The plate had one limit, 50 hours, for quests and curiosity alike. Nothing stopped curiosity filling more hours than the curiosity cap could ever count.

Now: A quest needs a commitment outside my head: a date someone else holds me to, a person waiting on it, or money already spent or at stake. Good for me someday does not count. The lane changes on an event: an exam I mean to sit is curiosity until the day it is booked, paid for or needed by someone. An exam's date counts toward the floor only once it is marked booked. A sitting I could still move is a plan, not a commitment. The two ICS exams are booked. SQE 1 and HTB CPTS are curiosity until they are. This week's curiosity cannot pass this week's curiosity cap in hours: 60 points is 15 hours this week, 20 at a full bar. This week's curiosity means the hours already logged plus what the plate still asks for, so a week that has had 14 hours of curiosity has room for one more, whatever the plate says. Anything that would push it over is refused, with the hours that still fit offered instead; the logged hours start again on Monday. Like the plate's own limit, only an increase is checked, so a plate that fit is never forced off when the cap drops. A lane change never reaches back. Work logged before it keeps the lane it was done in. Turning SQE 1 to curiosity first moved its October chapters out of the quest lane, which pushed the week of 26 October over its curiosity cap and dropped this week's bar from 150 to 145 in the middle of the week. Its file now records that it was a quest until 5 November, and the bar is 150 again. The floor change applies from Monday 9 November. The week that started on 2 November keeps the floor of 101 it began with. Next week's floor goes from 159 to 146.

2026-11-04

Practice boxes have a rate

Previously: Practice boxes (Hack The Box machines and the like) had no rate of their own. They could only go on the timer, at the half rate for work with no button, or into a guessed price for a whole certification. HTB CPTS was priced that way: 320 hours for the path and the boxes together, cut into 28 modules.

Now: A practice box is an activity on the menu, priced by the platform's own difficulty rating at a learner's pace, and done when root is owned: - Easy: 2 hours, 12 points. - Medium: 4 hours, 24 points. - Hard: 8 hours, 48 points. Each is times 1.5, since owning a box is retrieval: it is where the modules turn into skill. The rate is fixed, so getting faster pays. There is no set number of boxes; I do as many as I need until I am ready. HTB CPTS moves from Tasks to Credentials, so it works like any exam: booking it gives it a fixed date that counts toward the floor, and passing pays its outcome. It starts off the plate, as it was in the backlog. One correction, made once and before any of it was logged: since the boxes now pay on their own, the 320 hours no longer fits. The 28 modules are priced at 150 hours for the path alone, at the guess factor in force today (1.8): 1,080 points, and the same again on passing.

2026-11-04

Practice hours count before they are done

Previously: An exam's hours a week and its share of the floor came from its modules alone. The practice that actually passes it (flashcards, past papers, questions) paid as I logged it, but its time was invisible until it was done, unless I planned a count for each activity, and I do not think in counts. I think in hours.

Now: Something with a date can carry practice hours: how long all its activities together should take. What the logged activities have not yet covered counts toward its hours a week and, for a fixed date, the floor. Practice is measured by what I log, at each activity's own rate in real time (a past paper is its official length and half again to mark it), never by the clock on a timer. The floor counts it at four points an hour, without the retrieval weighting, so it errs low. Practice hours never pay; the activities do, as before. Practice hours replace the per-activity plans on an item that has both. From today, both ICS exams carry 25 hours of practice each, on top of their modules. Like any plan, it reaches the floor from the next Monday, 9 November.

2026-11-02

The clock on a session means little

Previously: When the monthly review matched time to the parts it bought, overlapping sessions counted once: a session only added the part of its span no earlier one covered. Time typed in by hand was placed by the clock times on its line.

Now: I pause timers, delete them and type time in afterwards all the time, so the clock times on a line say little. A typed-in line is stamped when I log it, not when I did the work. Overlaps between sessions mean nothing, and trimming them would quietly throw real minutes away. So every session counts its own minutes in full, overlaps or not. Time typed in by hand goes by its date: it counts toward every part finished that day, or the first one finished after it, when the review checks whether a part was partly typed in. This touches only the monthly review and the notes on parts that ran far over or under their price. Points, hours logged and the weekly bar never looked at the clock. On the record so far it changes nothing: the 2 November review comes out at the same 1.8.

2026-11-02

Planned work counts before it is done

Previously: Activities (past papers, practice questions, flashcard reviews) paid as I logged them, but had no finish line, so they never counted toward an item's hours a week or the bar's floor. Only an item's parts did. Once ICS Shipping Business was re-measured, its modules were worth almost nothing, and the system thought an exam sixteen days away needed twelve minutes a week. The work that actually decides the exam was invisible.

Now: An activity on something with a date can carry a plan: how many I mean to do in all. What is left of the plan counts toward the item's hours a week and, for a fixed date, toward the floor. The plan never pays. Only logged units do, at the same rates as before, so planning big buys nothing but a heavier week. A plan made during a week counts toward the floor from the next Monday, the same as everything else that sets a week's numbers. A re-measure works the same way now: a week that started before it keeps the floor it began with. From Monday 9 November. One correction, made once: the Shipping Business re-measure counted the same notes twice, as 54 pages and as 6,000 words. Its six modules are priced from the words alone, 2 points rather than 6. The re-measure form now asks for printed pages or words, not both for the same text.

2026-11-02

Every part pays

Previously: A group of parts shared its price, and each part paid its share, rounded. When a group was worth less than a point a part, some parts paid nothing. Shipping Business's last six modules shared 2 points, so they paid 0, 1, 0, 0, 1, 0. I finished a module, tapped the button and got nothing.

Now: Every part is worth at least one point. A button that finishes a piece of work never pays nothing; a group priced below a point a part is raised to one, and a group cannot be split into more parts than it has points. Fifteen minutes is the smallest piece of work that gets a button of its own. Pages are the exception, since a book is logged by the page I reach rather than a tap a page. Counted activities (cards, questions) also pay by count: a small batch can round to nothing until the next one carries it, because otherwise logging one card at a time would pay forty times the rate. Timed work pays by the hour, so a session too short to round to a point pays nothing. It applies from today, to everything priced from now on. The only group it touched was Shipping Business: its six modules are now 6 points, a point a module, and the one already finished pays its point. Nothing else is re-priced, and nothing earned before moves. The 24 points the first module earned at the old guessed price stay; a price is never cut later to make up for an earlier one.

2026-11-01

The bar moves with the work

Previously: The weekly bar was 200 points, fifty hours, every week. Curiosity could fill 80 of it, and the plate could hold fifty hours a week.

Now: A bar I miss week after week only teaches me to miss it, the same way ending on six of fourteen did. So from Monday 2 November the bar is set at the start of each week and moves with what I actually do: - It moves halfway from last week's bar to what last week counted, to the nearest five. Score 100 against 200 and next week's bar is 150. Beat it and it rises the same way. - It never drops below what work with a date someone else fixed needs that week: every exam, and any task whose date was set for me, whether it is on the plate or not. - It never goes above 200. - The curiosity cap is 40 percent of the bar. The plate keeps its fifty hours whatever the bar is. On the first day I tied the plate to the bar too, and a bar of 150 left room for only 37.5 hours, less than the work I actually had. The bar says when a week is full, not how much there is to do, so a low bar should never push work off the plate. Each week's numbers are fixed by what happened before it started, so swapping things on and off the plate midweek changes none of them. All of it is worked out from the record, with nothing to set by hand.

2026-11-01

No cap on a timed session

Previously: Time on work with no button of its own (labs, side reading, a one-off) scores at half rate, two points an hour. One session counted for at most three hours, on the theory that a long stretch on a clock always hides breaks, and a timer cannot tell focus from drift.

Now: The cap is gone. I already take the breaks out myself: at the end of a session I confirm the focused minutes, and they can only go down. A cap on top of that does not catch drift, it just stops paying for the fourth hour of a long session that was focused. Half rate stays, because hours are still input, not output. From Monday 2 November, every focused minute counts. Sessions before then keep the points they were frozen at. One exception, made once and by choice: the five hours of labs on 1 November that prompted this are re-scored without the cap, at ten points rather than six.

2026-10-27

Prices come from the work, not from my speed

Previously: Every point on the weekly bar is priced in advance from an estimate of how long the work will take. One calibration factor corrected every new estimate: all the timed work on finished parts in the last eight weeks, set against what those parts were priced at, across everything at once. Once five parts were in, that single ratio multiplied every new guess and every new formula.

Now: On 27 October that factor read 0.5, which would have halved every new price. It was an artifact. One ten-hour course section carried almost three quarters of the weight, so five finished parts held the information of fewer than two, and that one section was about to halve exam prices that had come within about 13 percent of the time they took. The section's own price had been extrapolated from one earlier section, and the second section took less than a third as long as the first. One part was speaking for a course, and one course was speaking for every kind of estimate. So the single factor is gone, and the 0.5 with it, undone as a calculation error. Prices now come from something I cannot move by working slowly: an official timing if there is one, then the content counted (pages, words, video), then published hours, and only then my own guess, written down raw. Each of those has its own factor, and each learns only from its own record: once a month, on the first Monday, from items rather than minutes, with no item worth more than a quarter of the evidence, and never more than a tenth at a time. My speed never re-prices anything. The one exception is a single re-measure from the content after the first part of something priced from published hours or a guess, made whatever that part took. No part is worth more than 16 points, or 8 for a guess, so one bad estimate can only be so wrong. Only time run on the timer counts toward any of it. One exception, made once and by choice: the first two sections of the course that exposed all this were priced before these rules, the second at 40 points from one recalled section, and it took under three hours. Both are re-scored from their content by the rules above: the first, finished on 19 October, at 28 points, and the second at 11. Points already earned are otherwise never rewritten. These rules start today. Any change from here goes in this log first and applies from the Monday after, and every proposal has to pass one test: an extra minute on the clock must never buy a minute or more of future price.

2026-08-16

The notebook should be one thing

Previously: The notebook had three shapes. Essays for the long arguments, notes for the short observations, and threads to gather either kind under a named line of thinking. The brief called the thread map the foreground and the reverse-chronological feed mere reference, and I believed that. The thread map was going to be the receipt of my reasoning: proof that I was working on something over time rather than posting into the void.

Now: Three shapes meant nothing got written. Fifteen months of that structure produced ten essays, nine of which were tweets I had already written elsewhere, and one genuine note. Every time I had something to say I first had to decide what kind of thing it was and which thread it belonged to, and often enough the answer was that I would think about it later. The filing system was not organising the writing. It was standing in front of it. So the notebook is essays now, and only essays. One kind, one list, newest first. Threads are gone; wikilinks and backlinks already do the job of showing which pieces connect, and they do it without me maintaining a taxonomy by hand. Drafts are the only remaining state, and a draft is a boolean, not a category. The nine tweets survive as drafts. They are honest artifacts of being eighteen and paying attention, but they are not essays, and I would rather the public list be eleven things I would defend than twenty things I have to explain. What I got wrong was mistaking a structure for a practice. The thread map was a promise about how I would think, made in advance of doing the thinking. The correction is to build only what the writing actually demands, and to add structure back when a real piece of writing asks for it, not before.