Changes and corrections — Redmoon Calculators

Every calculator on this site is a small program, and small programs have bugs. The ones that matter hand you a confident number that is wrong: a royalty that ignores a fee, a bag count that comes up short on delivery day. We look for them by reading each tool's code against the rule or formula it claims to follow and re-working its output by hand. When we find one it is listed here with the old number and the new one, and any guide that described the old behaviour is corrected the same day and shows an "Updated" date. If you find one we missed, the contact page explains what to send.

October 13, 2026: 16 calculators fixed, 14 tool pages and 23 guides corrected

The audit of 12 October left a list of problems it had found but not yet fixed. These are those fixes. Where the right answer depended on a published method or an operator's rules, we went back to the source: Smith and Senter's 1967 report for the Automated Readability Index, the digitised grade lines of Fry's 1977 graph, Rachel McAlpine's EFLAW bands, X's twitter-text counting rules, ACX's royalty help pages, Mean Well's supply ratings, York Wallcoverings's trim allowance, and the EIA, AAA and fueleconomy.gov for the EV defaults. Where it was a judgement call, the tool page now says which convention the tool follows.

Calculators that gave a wrong answer or behaved differently from their description

Tool	What was wrong	What it does now
ACX Audiobook Royalty Calculator	Applied the narrator royalty share to non-exclusive titles too, although ACX only offers Royalty Share (and Royalty Share Plus) to titles distributed exclusively: a non-exclusive $19.95 title with a 50% narrator share showed $2.99 a sale to the author ($29.92 a month at 10 sales).	With non-exclusive selected the narrator share is zero and the field explains why (a non-exclusive narrator is paid per finished hour), so the same title shows $5.98 a sale and $59.85 a month. Exclusive titles are unchanged.
Automated Readability Index (ARI) Calculator	Counted every non-space character, punctuation included, as an ARI character, about half a grade too high on ordinary prose: the page's example sentence scored 9.1.	Counts letters and digits only, as in Smith and Senter's 1967 report (AMRL-TR-66-220): the same sentence scores 8.8 (69 characters over 14 words instead of 70). The Readability Consensus Score uses the same ARI figure.
Syllable Counter & Analyzer	Skipped the silent-e rule on three-letter words, so "use", "one", "age" and "ice" counted 2 syllables. Every syllable-based score inherited it.	They count 1. The silent-e rule now applies from three letters, with "see", "tie" and "toe" still protected. Scores move by a fraction of a grade on most prose.
Fry Readability Graph Calculator	Estimated the grade with a straight-line blend of the two coordinates instead of Fry's graph, which read two to three grades high mid-chart: Fry's own worked example (141 syllables, 6.3 sentences per 100 words, grade 7) came out as 10.5. It also used only the first 100 words, counted numerals by vowels and never flagged the graph's invalid areas.	Looks the point up on the grade lines of Fry's 1977 graph (digitised by Readability Studio) on its uneven sentence scale, so Fry's example reads grade 7. It averages three 100-word samples (start, middle, end) when there are 300 words or more, counts numerals and initialisms one syllable per symbol as Fry specifies, needs at least 100 words, and labels points in the long-sentence and long-word areas as outside the valid area.
McAlpine EFLAW Readability	Used band edges that did not match McAlpine's published bands (1–20, 21–25, 26–29, 30+): a score of 25.0 to 25.4 was rated moderate and 29.0 to 29.4 difficult. One FAQ said short words are what learners find easiest, the opposite of the formula.	Rounds the score and applies McAlpine's bands and names: 25.4 is "quite easy to understand", 29.4 "a little difficult", 30 and up "very confusing". The FAQ now says short idiomatic words are what the formula counts as harder.
SEO Meta Tag Length Checker	Measured titles at 16px, against 580px desktop and 520px mobile limits, with a five-width character table, while Google sets titles at about 20px and other checkers use 20px against about 600px. The page's example title (44 characters) measured 367px; a 110-character title measured 943px. Descriptions used 13px against 960/860px.	Uses Arial's own character widths. Titles are measured at 20px against 600px on both devices (mobile wraps long titles rather than cutting them): the example title is 422px and the long title 1,070px. Descriptions are measured at 14px against 920px desktop and 680px mobile, the figures most SERP-preview tools use.
Tweet / X Length Checker	Counted a link as 23 only when it started with http:// or https://, so "Visit x.com" weighed 11 although X links bare domains.	Follows twitter-text: a domain without a scheme counts as a 23-character link when it ends in a real top-level domain, using twitter-text's own TLD list, so "Visit x.com" weighs 29; "@user.com", "#tag.io", email addresses and invalid endings such as ".comm" stay text.
Sentence Complexity Analyzer	Had a Mode menu (Plain English, Academic, ESL) that changed nothing: every choice flagged sentences over 35 words or with 3+ subordinators.	The menu is now an audience setting that changes what is flagged: plain English (default) flags sentences over 25 words, GOV.UK's limit; ESL over 20 words (the Simplified Technical English cap) or 2+ subordinators; academic over 35 words. The clause-depth rating bands move with it.
Vocabulary Sophistication Analyzer	Had Dictionary (American/British) and Mode (General/Academic/ESL) menus that changed nothing. British spellings such as "colour" and "centre" counted as advanced because the familiar-word list is American, and a text at 40% advanced was rated amber while the card called 20–45% ideal.	British spellings of familiar words now count as everyday, so the Dictionary menu was removed. Mode is now an audience setting with its own target band: general readers up to 45% advanced (so 40% is now green), learners up to 30%, academic readers up to 60%.
Text Cohesion & Transition Analyzer	Showed a Mode menu (Essay, Blog, Technical) that changed nothing.	The menu is removed; the score and its 60+ guidance were always the same for every choice.
Text Formality Analyzer	Missed contractions written with a typographic apostrophe, so "we’re" and "don’t" pasted from a phone or word processor counted 0 contractions.	Contractions are matched with either apostrophe, so they count the same as typed text.
Wallpaper Calculator with Pattern Repeat	Took one total wall width and cut every drop at exactly the wall height, with no trim allowance, so a 12 × 8 ft room (480 in of wall, 96 in high, 33 ft rolls) got 4 drops a roll, 24 strips and 7 rolls.	Takes each wall's width, counts drops wall by wall and adds a trim allowance (4 in by default, York Wallcoverings's figure) to every drop, with a straight or half-drop match: the same room needs 26 drops at 3 a roll, 10 rolls (24 drops if every corner offcut could be reused, which the result also shows).
LED Strip Power Supply Calculator	Picked supplies from 12, 20, 30, 60, 100, 150, 200 and 350W, so a 205.9W requirement jumped to 350W, and warned about voltage drop on a 16.4 ft (5 m) reel at 12V, the length 12V strip is made to be fed from one end.	Picks from the standard ratings of Mean Well's LED driver and supply ranges, 20W to 600W (205.9W gets 240W; 63.4W gets 75W), shows how loaded the supply is, and warns only past one 5 m reel at 12V or 10 m at 24V.
Fantasy Travel Time Calculator	The terrain menu gave Mountain/swamp and Desert the same value, so after choosing Mountain the menu showed Desert (the speed was right; the label was wrong).	Each terrain has its own entry, so the menu shows the one you chose. Both still multiply speed by 0.5.
EV Charging Cost & Road Trip Calculator	Defaulted to $0.16/kWh, 12% charging losses and $3.50/gal gas, none checked against current figures; a full 75 kWh charge showed $13.44 and the 300-mile trip saved $19.64 against gas.	Defaults are $0.18/kWh (EIA, US residential average 18.31¢ in July 2026), 10% losses (fueleconomy.gov) and $4.35/gal (AAA national average, 12 October 2026): $14.85 a charge and $26.53 saved on the trip. The page now says to set losses to 0 when entering an EPA rating, which already includes them.
Craft Fair Profit Calculator	Had no field for costs such as parking, lodging or meals, although the page told users to count them.	An "Other event costs" field (default $0) is added to total expenses; the default results are unchanged.

Tool pages whose explanation did not match the tool

Mostly worked examples whose numbers no longer matched what the calculator shows for the same inputs, and explanations that described features a tool does not have. Each page now matches its tool.

Automated Readability Index (ARI) Calculator (2 corrections)

The worked example said the tool counts every non-space character, punctuation included, and gave 9.1.
How-it-works did not say which characters count.

Fry Readability Graph Calculator (2 corrections)

The worked example gave grade 8 for 140 syllables and 6.5 sentences; on Fry's graph that point is grade 7.
The explanation said the grade was an approximation of Fry's curves that does not flag the invalid areas, and that the tool reads only the first 100 words.

Readability Consensus Score (1 correction)

The in-depth section described the consensus as an average; it is the median of the nine grades.

McAlpine EFLAW Readability (3 corrections)

How-it-works gave the bands as ≤20, 20–24, 25–28 and 29+.
An FAQ said short words are what ESL readers find easiest; EFLAW counts them as harder.
The worked example called 24 "easy"; McAlpine's band name is "quite easy to understand".

SEO Meta Tag Length Checker (2 corrections)

The worked example gave the title as 367px against 580px and the description as 267px against 960px.
The FAQ and in-depth section gave 580/520px title and 960/860px description limits.

Tweet / X Length Checker (1 correction)

How-it-works said only links starting with http:// or https:// count as 23.

Sentence Complexity Analyzer (1 correction)

How-it-works, an FAQ and a pitfall gave a single 35-word split rule and did not mention the Mode menu, which did nothing.

Vocabulary Sophistication Analyzer (1 correction)

How-it-works did not mention the Dictionary and Mode menus, which did nothing.

ACX Audiobook Royalty Calculator (3 corrections)

The in-depth section still said exclusive pays 40% and non-exclusive 25%, through Audible, Amazon and iTunes.
An FAQ called exclusive "Audible-only" and named Apple Books as a non-exclusive outlet; exclusive titles are sold on Audible, Amazon and Apple Books.
Nothing said Royalty Share needs exclusive distribution.

Craft Fair Profit Calculator (2 corrections)

The worked example used the old $0.67/mile rate ($406.20 profit, $40.62/hr); at the tool's $0.76 default it is $402.60 and $40.26/hr.
The in-depth section listed parking and accommodation as costs the tool had no field for.

Dice Bag Capacity & Weight Estimator (2 corrections)

The in-depth section was about drawing tokens from a bag without replacement, not about this calculator, which estimates the volume and weight of dice sets.
The tool description said it accepts individual d4–d20 counts; it takes whole 7-piece sets.

Wallpaper Calculator with Pattern Repeat (2 corrections)

The worked example (7 rolls) and the FAQ on measuring told users to enter one total wall width with no trim allowance.
The in-depth section said a half-drop match always wastes less than a straight match; it depends on the wall height.

LED Strip Power Supply Calculator (1 correction)

The worked example said one 5 m reel at 12V needs power injection; the FAQ listed the old 12–350W size steps.

EV Charging Cost & Road Trip Calculator (2 corrections)

A pitfall said using a fast-charging rate makes costs look artificially low; it makes them look high if you charge mostly at home.
The FAQ and in-depth section gave charging losses as 10–15% or 10–20% without a source and did not say that EPA ratings already include them.

Guides corrected

Each now shows an "Updated" date under its title that links back here.

The Gunning Fog Index, explained: what the score means and how to lower it (1 correction)

Said the readability consensus averages several formulas; it takes their median.

The SMOG index: the readability formula health writers actually trust (1 correction)

Said the readability consensus averages SMOG with the others; it takes their median.

The LIX readability index: the formula that does not care about syllables (1 correction)

Said the readability consensus averages several formulas; it takes their median.

Dale–Chall readability explained: the formula that scores words you actually know (1 correction)

Said the readability consensus averages several formulas; it takes their median.

The Automated Readability Index: why counting characters beats counting syllables (3 corrections)

Said the ARI calculator counts every non-whitespace character, punctuation included; it now counts letters and digits, as the published method does.
Said punctuation inflates the character count; only digits and long tokens such as URLs do now.
Said the readability consensus averages several formulas; it takes their median.

Linsear Write: the readability formula that counts easy and hard words by hand (1 correction)

Said the readability consensus averages several formulas; it takes their median.

FORCAST: the readability formula that ignores sentences entirely (1 correction)

Said the readability consensus averages several formulas; it takes their median.

How to read the Fry Readability Graph (and why it still beats a single grade number) (2 corrections)

Said the calculator samples only the first 100 words and approximates Fry's curves, differing from a hand plot by a grade or two; it now averages three samples and reads the grade from the graph's own lines.
Said the tool does not flag the invalid areas; it now does.

Why your 280-character tweet is actually too long: weighted counting explained (1 correction)

Said the checker counts a bare domain such as x.com letter by letter; it now counts it as a 23-character link, as X does.

EFLAW: the readability score built for non-native English readers (1 correction)

Gave the bands as 20 or below, 21–24, 25–28 and 29+; McAlpine's are 1–20, 21–25, 26–29 and 30+.

Why your meta tags get cut off: pixels, not characters (1 correction)

Gave 580px desktop and 520px mobile title limits and 960/860px description limits, and described the old five-width character table; the checker now measures titles in 20px Arial against 600px and descriptions in 14px Arial against 920/680px.

The pre-publish check for a how-to blog post: readability, keywords, meta tags and reading time (2 corrections)

The title and description widths came from the old 16px measurement: the draft title is 1,070px (was 943), the final title 545px (was 480), and the final description is long on mobile under the new 680px limit.
Recomputed after the ARI and syllable fixes: draft ARI 24.3 (was 24.7) and edited 6.2 (was 6.7), Flesch–Kincaid 21.4 (was 21.5), FORCAST 12.5 (was 12.6), Flesch Reading Ease 14.1 (was 13.8), spread 34.0 (was 33.9), first paragraph Flesch −9.5 (was −10.3). The consensus grades (20.7 and 8.0) are unchanged.

Clause depth, not word count: how a sentence complexity analyzer reads your syntax (1 correction)

Gave a single 35-word split rule; the analyzer now uses 25 words for plain English, 20 for ESL readers and 35 for academic writing.

Measuring vocabulary sophistication: rare words, CEFR levels, and when "advanced" hurts you (1 correction)

Said the analyzer turns yellow past about 35% advanced; for general readers it is now amber past 45% and red past 60%, with tighter bands for learners and looser ones for academic readers.

Is that craft fair booth actually worth it? Do the math first (1 correction)

Listed the calculator's inputs without the new other-event-costs field.

What it actually costs to charge an EV vs. fill up with gas (2 corrections)

Said DC fast charging tends to carry higher conversion and thermal losses than home charging; measured from the grid the two overlap (ADAC: 5–15% DC, 5–7% on an 11 kW wallbox), and the cost gap comes from the price per kWh.
Used $0.16/kWh, 12% losses and $3.50 gas; with current figures ($0.18, 10%, $4.35) a full charge is $14.85 and the trip saves $26.53, and $0.40/kWh fast charging no longer costs more than gas (break-even about $0.46).

Why your EV draws more kWh than its battery holds — charging losses explained (3 corrections)

Said home Level 2 losses are commonly 10–15%; fueleconomy.gov uses about 10% and wallbox tests measure 5–10%.
Did not say that EPA efficiency ratings already include charging losses, so adding a loss percentage to an EPA figure counts them twice.
Said a fast-charged car can cost several times more per mile; at typical prices it is about two to three times.

Sizing an LED strip power supply (and avoiding voltage drop) (2 corrections)

Listed supply sizes 12–350W with a 101W requirement jumping to 150W; standard sizes include 120W.
Said a single 16.4 ft (5 m) reel at 12V is past the single-feed limit and needs injection; one reel is what 12V strip is made to be fed with from one end.

Wiring several LED strip runs to one power supply (1 correction)

Gave the 12V single-feed limit as about 16 ft and the size list as 12–350W.

Wallpaper math: why pattern repeat changes how many rolls you need (1 correction)

The worked example cut drops at exactly the wall height with no trim and summed all walls into one width (7 and 9 rolls); with a 4-inch trim and wall-by-wall counting the example is reworked on a 9 ft room (10 and 15 rolls, 10 with a half-drop match).

Dye lots, deductions, and the waste buffer: ordering wallpaper you won't have to reorder (1 correction)

Told readers to enter the sum of all wall widths and to leave trimming to the waste buffer; the calculator now takes each wall and a separate trim allowance.

Exclusive vs. non-exclusive ACX distribution: the real royalty math (2 corrections)

The description and closing line still quoted the legacy 40% and 25% rates.
The tool description did not say the narrator share applies only to exclusive titles.

One 90,000-word novel, three formats: from draft plan to Kindle, paperback and audiobook royalties (1 correction)

Showed a non-exclusive royalty-share row ($2.70 a sale to the author); ACX requires exclusive distribution for royalty share, so a non-exclusive title pays its narrator up front and keeps $5.40 a sale ($161.91 a month).

October 12, 2026: 29 calculators fixed, 78 tool pages and 57 guides corrected

Writing guides that chain several calculators meant re-running every tool those guides use and checking it against its page and its older guides. We then did the same for every other tool and guide on the site. These are the results. Where a fee schedule or published rule had changed (Amazon KDP's royalty rules, ACX's royalty rates, the IRS mileage rate), the calculator was updated to the operator's current published figures.

Calculators that gave a wrong answer or behaved differently from their description

Tool	What was wrong	What it does now
Wargame Match Time Predictor	Estimated a 2,000-point Warhammer 40,000 game between two casual players at 50 minutes, a third of a realistic length; its own worked example said about 150 minutes, and the other systems were off by similar factors.	Per-point rates are recalibrated so a standard-size casual game lands near two to two and a half hours: the same 2,000-point game is now 150 minutes, 1,000 points of Star Wars: Legion 150 minutes and 1,000 points of Bolt Action 120 minutes.
Miniature Scale & Terrain Converter	Rounded the foam-layer count up from a floating-point result, so an exact fit could report an extra sheet: a 15 ft wall at 1:60 on 12.7 mm sheets (exactly 76.2 mm) showed 7 layers.	The same wall now shows 6 layers.
Sci-Fi Comms & Orbital Delay Calculator	The result sentence said Earth would wait the one-way delay (4 min 19 s at Mars's closest) for the outpost to acknowledge a message; an acknowledgement needs the round trip.	It now gives both: the message arrives after 4 min 19 s and the earliest acknowledgement after 8 min 39 s.
Syllable Counter & Analyzer	Counted one syllable too many in words ending in a consonant plus "le" ("table", "simple" and "people" scored 3, "example" 4) and did not treat vowel plus "le" as a silent e ("rule" and "while" scored 2). Every readability formula that counts syllables inherited the error, and "simple" was counted as a complex word by Gunning Fog and SMOG.	"table" is 2, "example" 3, "rule" 1. Grade-level scores on ordinary prose come down slightly, mostly by a fraction of a grade.
Dale–Chall Readability Calculator	Displayed the raw Dale–Chall score as a US grade, so a raw 9.5 (college level on the formula's own table) was labelled "grade 9.5, fairly difficult". Regular inflections such as "improved" and "cities" were also counted as unfamiliar although their roots are on the list.	The result is shown as a raw score with the formula's published bands (for example 9.0–9.9 is "College (grades 13–15)"), and inflected forms match their roots.
Readability Consensus Score	Fed the raw Dale–Chall score into the median of grade levels as if it were a grade.	Dale–Chall now enters as the grade at the middle of its band (raw 7.3 counts as grade 9.5).
KDP Royalty & Book Price Calculator	The 70% ebook royalty subtracted the full delivery fee after taking 70% of price, e.g. $2.59 for a $4.99, 6 MB book. The ebook royalty card described the royalty as 70% minus the delivery fee. Paperback printing cost used $0.85 + $0.012 per page beyond 108, giving $3.15 for a 300-page book. Paperback royalty was 60% of (price − printing cost) with no 50% band, e.g. $7.10 at $14.99 for 300 pages and a $5.26 floor. The paperback royalty card described the royalty as 60% of (price − printing cost). The 70% tier option said the eligible price band is $2.99–$9.99. The ebook result note said 70% needs a list price between $2.99 and $9.99.	Delivery cost is now deducted before the 70% rate as KDP specifies, so the same book shows $2.86. It now says 70% of (price − delivery fee), matching KDP's formula. It now uses KDP's US black-ink table ($2.30 up to 110 pages, else $1.00 + $0.012 per page), giving $4.60. Royalty is now rate × price − printing cost with 50% below $9.99, giving $4.39 at $14.99 (after the print-cost fix) and a $9.20 floor. It now reads 60% of price − printing cost, with the 50% rate under $9.99 noted. It now says $2.99–$12.99, the Amazon.com band since 7 July 2026. It now says $2.99–$12.99 on Amazon.com.
ACX Audiobook Royalty Calculator	ACX royalties used the legacy 40% exclusive / 25% non-exclusive rates, e.g. $7.20 a sale at $17.99. The exclusive option label said 40%. The non-exclusive option label said 25%. The royalty card subtitle showed 40% / 25% of price.	It now uses the new-model 50% / 30% rates that all titles move to by end of 2026, e.g. $9.00 a sale at $17.99. It now says 50%. It now says 30%. It now shows 50% / 30%.
Craft Fair Profit Calculator	Craft Fair Profit Calculator defaulted the mileage rate to $0.67/mile, the 2024 IRS business rate.	Default mileage rate is now $0.76/mile, the IRS business rate for July-December 2026.
Laser Cutter Job Cost & Margin Calculator	The laser cutter calculator labelled a 100% markup on cost as '+100% margin'; the real margin at the default is 50%.	The retail card now reads '+100% markup', matching the formula.
3D Print Cost & Profit Estimator	The 3D print calculator labelled a 50% markup on cost as '+50% margin' (true margin 33.3%).	The retail card now reads '+50% markup'.
Resin & Epoxy Cost Calculator	The resin calculator's volume caption said '24.0 in³ incl. 3% waste', but 24.0 in³ is the mold volume before waste.	The caption now reads '24.0 in³ + 3% waste'; gallons, ounces and cost were already correct.
Epoxy Garage Floor Coating Calculator	The epoxy flooring calculator showed a 'Coverage / gal / coat' field that had no effect on any result (changing 175 to 50 left 4 kits and $708.40 unchanged).	The unused field is removed; the kit count is driven by the kit coverage field, as before.
Herringbone & Chevron Tile Calculator	Floating-point rounding made the tile calculator add a phantom tile in some cases: a 12 x 10 ft straight lay with 24 x 4 in tile showed 199 tiles instead of 198. With straight lay selected, the tile calculator said the 10% extra was 'for angled cuts'.	The count is rounded to 6 decimals before rounding up, so that case now shows 198 tiles (20 boxes, unchanged). Straight lay now says the 10% covers edge cuts and breakage; herringbone and chevron keep 'angled cuts'.
Board & Batten Wall Calculator	With a single batten the board and batten summary read '1 battens'.	It now reads '1 batten'.
Indie Game Audio File Size Estimator	Audio estimator labeled its MP3 option 192 kbps but sized it at about 113 kbps (20 min: 16.1 MB instead of about 27.5 MB). Audio estimator made OGG, MP3 and Opus files 50% bigger when the source was 24-bit (20 min OGG: 21.2 MB vs 14.1 MB).	The MP3 option is now labeled with the bitrate it actually models, about 113 kbps. Compressed sizes no longer depend on source bit depth (20 min OGG: 14.1 MB either way).
NAS / RAID Usable Capacity Calculator	RAID calculator's overhead hint told users to include the TB/TiB gap, which it then applied a second time. RAID calculator treated RAID 1 with more than two drives as mirrored pairs (5 × 8 TB: 14.4 TB usable, 2 failures). RAID calculator showed RAID 10 fault tolerance as drives/2 − 1 (5 drives: 1.5, 8 drives: 3).	Hint now asks only for filesystem metadata and reserved space. RAID 1 is now an n-way mirror as in Synology and Linux (5 × 8 TB: 7.2 TB after 10% overhead, 4 failures). RAID 10 now shows the guaranteed tolerance of 1 drive.
Catering & Party Food Quantity Calculator	Catering calculator headed served (cooked) weights 'Shopping list', contradicting its own page note that they are not purchase weights.	The meal results are now headed 'Served quantities'.
Raygor Readability Graph Calculator	The Raygor calculator counted only words of seven or more letters as long, although Raygor's method (and our own Raygor guide and page) counts six or more.	The Raygor calculator now counts words of six or more letters as long, as the published method does, so six-letter words such as 'simple' and 'though' raise the estimate.
Cliché Detector	The cliché detector counted one cliché two or three times when the list held overlapping entries: "the home stretch" counted as 2, "when all is said and done" as 3, so a sentence with 3 clichés reported 7 and inflated the per-100-words density.	Each stretch of text now counts once, under the longest matching phrase: the same sentence reports 3 clichés.
Filler / Weak Word Detector	The filler detector filed "quite" and "rather" under qualifiers, although its own list and the filler guide put them under intensifiers (the duplicate entry overwrote the intensifier one).	"quite" and "rather" are now counted as intensifiers, matching the guide.
Tweet / X Length Checker	The tweet length checker counted Hebrew and Arabic letters as 2 each and every emoji code point as 1 (a family emoji cost 5, a single emoji 1), the opposite of X's published twitter-text v3 rules.	Hebrew, Arabic and everything else up to U+10FF now count 1; each emoji (including ZWJ, skin-tone, flag and keycap sequences) counts 2; CJK, Hangul and other code points count 2, matching twitter-text v3.
Lexical Diversity (TTR) Calculator	The lexical diversity tool showed only the number of hapax legomena, while the post and tool page told readers to skim the list of them.	The tool now lists up to 200 words used exactly once under the frequency table, as the post and page describe.
AI Text Pattern Detector	The AI pattern detector's hedge list included 'tend to' and 'tends to' but they were never counted (single-token comparison), so hedge rate missed them; the post lists 'tend to' as a hedge the tool measures.	Multi-word hedges are now matched on the full text and added to the hedge count.
Scattergories Generator	The Scattergories bank listed 'Things made of glass' twice (easy and medium), so mixed rounds could show it twice.	The medium duplicate is now 'Things made of rubber'; every category text is unique and the bank stays at 322 (medium 132).
Comic Script Page-to-Panel Pacing Calculator	Comic pacing tool printed the page-total dialogue/action rating under the per-panel figure (e.g. '6.7 words/panel — cramped')	The rating now names the page total it was judged on (e.g. 'cramped (40 words/page)')
Cosplay Fabric & Material Estimator	Cosplay estimator's on-screen note said it uses π × radius² while it computes 2 × π × radius²	Note now says the circle area is doubled to allow for offcuts
Conlang Syllable & Phonotactics Checker	Conlang checker rejected any word ending in a consonant under CV/CVC rules (e.g. 'kat', 'manta'), because it split syllables greedily and never tried CVC after CV; its own generated words failed its own check	The checker now tries every way of splitting a word into the allowed syllable shapes, so 'kat', 'manta' and 'monel' pass under CV/CVC while 'kthaxel' is still flagged at position 1
TCG Booster Pack Drop-Rate Calculator	TCG pack calculator had a 'Target copies' box that changed nothing: 4 copies of a 1-in-30 rare showed the same 89 packs for 95% as one copy TCG result sentence always said 'a specific rare card' whatever the copy target	Target copies now counts: 4 copies of a 1-in-30 rare need 110 packs for 50% and 231 for 95%; one copy is unchanged (21 / 89) Sentence now names the number of copies when it is more than one

Tool pages whose explanation did not match the tool

Mostly worked examples whose numbers no longer matched what the calculator shows for the same inputs, and explanations that described features a tool does not have. Each page now matches its tool.

Dice Pool Probability Calculator (1 correction)

The description promised exploding dice and mixed pools such as 3d6 + 1d4 + 2, which the calculator does not support, and the worked example quoted 78.2% for 3d6+5 against 15 (the tool gives 62.5%).

Miniature Scale & Terrain Converter (1 correction)

An FAQ said to divide by 56 for 28mm while the tool's preset divides by 60.

Wargame Match Time Predictor (1 correction)

The how-it-works and FAQ text described the old rates as "community survey averages".

Fantasy Travel Time Calculator (1 correction)

The worked example said 100 miles through forest takes 4 days 4 hours with 16 rations; the tool gives 5 days 4.4 hours and 24 rations for those inputs.

TTRPG Encounter Combat Pacing Predictor (1 correction)

The worked example used 70 seconds a turn and 37 minutes; the tool's formula gives 90 seconds and 48 minutes.

Miniature Paint Mix & Volume Calculator (1 correction)

The worked example gave 5 drops of thinner; each part is rounded up, so the tool shows 6.

TTRPG Settlement Demographics Calculator (1 correction)

The worked example gave a 5,000-person town a 1,000 gp limit and 250 guards (the tool shows 5,000 gp and 200), and the explanations said prosperity adds alchemists and more farmers, which it does not.

Dale–Chall Readability Calculator (2 corrections)

The worked example gave 11.4 with six of seven words unfamiliar; the sentence has eight words, five unfamiliar, and scores 13.9.
The Dale–Chall page used 'knowledge' as a familiar word, but the calculator flags it.

KDP Royalty & Book Price Calculator (8 corrections)

The worked example subtracted the delivery fee after taking 70% ($3.04); KDP takes it off the price first ($3.18).
The FAQ said the 70% band is $2.99 to $9.99.
The FAQ said paperback royalty is 60% of (list price − printing cost).
The depth copy said the 70% band is broadly $2.99 to $9.99.
The cliff example compared $9.99 at 70% with $10.99 at 35%, but $10.99 now earns 70%.
How-it-works said royalty = price × 0.70 − delivery fee.
How-it-works gave print cost as $0.85 + $0.012 per page beyond 108 and royalty as 60% × (price − printing cost).
How-it-works gave the floor as printing cost ÷ 0.60 only.

ACX Audiobook Royalty Calculator (5 corrections)

The worked example used the 40% exclusive rate ($7.98 a sale); under ACX's current model it is 50% ($9.97).
How-it-works used 40% / 25% of retail price.
The FAQ said exclusive pays 40%.
The FAQ said non-exclusive pays 25%.
The FAQ said royalty-share splits last for the life of the audiobook.

3D Print Cost & Profit Estimator (8 corrections)

3D print estimator example output said Raw $4.78 and Retail $7.17.
How-it-works said 3D print labor is computed from prep/post-process minutes; the tool has no minutes input.
FAQ said the 3D print estimator adds a wear allowance; it has none.
Depth copy said filament is often the smallest cost on a 12-hour, 80 g print; at the tool defaults filament ($2.00) is the second largest.
3D print calculator description promised a wear allowance and a tiered retail price
3D print calculator tagline listed machine wear, which it does not model
3D print page said a 10% buffer exactly covers a 10% failure rate
3D print page heading said filament is the smallest cost, while the paragraph under it shows electricity at 22 cents against $2.00 of filament

Etsy Fee & Profit Calculator (4 corrections)

Etsy fee example claimed fees ~$3.16, net ~$14.84 (49.5%), ~$296.80/mo.
Etsy fee pitfall said sales tax does not affect the seller payout; Etsy charges its 3% processing fee on the tax too.
Etsy depth copy said free shipping and charged shipping are not fee-equivalent; both are charged the same fees.
Etsy fee FAQ said shops over $10k a year 'may be enrolled' in Offsite Ads at 12–15%

Print-on-Demand Profit Calculator (2 corrections)

POD example claimed fees ~$2.88 and profit ~$13.10 (43.7%).
POD calculator page did not mention that it omits Etsy’s $0.20 per-sale listing renewal.

Candle Making Cost Calculator (2 corrections)

Candle example claimed unit cost ≈ $3.61 and price ≈ $9.92.
Candle depth copy said labor minutes are per batch and divided across 24 candles (about 16 cents each); the tool charges them per candle ($3.75 each).

Laser Cutter Job Cost & Margin Calculator (3 corrections)

The laser cutter page's worked example said raw cost $11.62 and retail $23.24.
The laser example listed power at $0.04.
The laser FAQ suggested retail at 2x cost with wholesale at 1.5x cost; the calculator's 60%-of-retail rule gives 1.2x at that markup.

Freelance Day Rate & Hourly Rate Calculator (2 corrections)

The freelance rate page example said $97.94 an hour and $783.52 a day.
The example explanation repeated $97.94 / $783.52.

Art Commission Pricing Calculator (1 correction)

The commission calculator page said it 'works backwards from the income you need, the hours you can actually bill and your overheads', but it only takes an hourly rate.

Resin & Epoxy Cost Calculator (3 corrections)

The resin casting calculator's explainer was written about resin 3D printing (vat film, IPA wash, supports).
The first paragraphs discussed IPA, FEP film and print post-processing, none of which the casting calculator models.
The second section opened with resin-print support material.

Epoxy Garage Floor Coating Calculator (1 correction)

The epoxy calculator's worked example listed a 175 sq ft/gal coverage input that the tool never used.

French Drain Calculator (3 corrections)

The French drain worked example said the total was about $243.94; the calculator shows $243.93.
The French drain explanation said gravel cost $158.94 and the total $243.94.
The French drain FAQ said an 18-24 inch trench puts the pipe below the frost line in most climates and below typical foundation footings.

Board & Batten Wall Calculator (1 correction)

The board and batten FAQ described the 14 inch default as on-center spacing, but the tool's input is the clear gap between battens (16.5 inches on center at the default).

LED Strip Power Supply Calculator (1 correction)

The LED supply page said the default 20% headroom equals sizing to 80% of rated output; 20% headroom actually loads the supply to about 83%.

Sprite Sheet & VRAM Budget Calculator (3 corrections)

Sprite sheet example output said a 128×64 sheet (4 × 2) using 32 KB.
Sprite sheet example explanation described a 4 × 2, 32 KB layout the tool never produces.
Sprite sheet page said 12 fps does not divide evenly into 60 Hz.

Retro Game ROM / RAM Budget Calculator (4 corrections)

Retro ROM example explanation summed to 35,392 bytes.
Retro ROM page called 32 KB / 128 KB / 256 KB 'the real' cartridge sizes for Game Boy, NES and Game Boy Color.
How-it-works said each system has a fixed ROM capacity (Game Boy 32 KB, NES 128 KB...).
Retro ROM FAQ described a bank-boundary warning the tool does not have.

Indie Game Audio File Size Estimator (4 corrections)

Game audio example said ~22 MB (~317 MB WAV) while the tool shows 21.2 MB (302.8 MB WAV).
Game audio pitfall said the format ratios are for 192 kbps; they correspond to about 99-113 kbps.
Game audio FAQ said speech sounds noticeably duller at 22 kHz, contradicting the next FAQ and the post.
Game audio page said any rate above 44.1 kHz doubles storage.

NAS / RAID Usable Capacity Calculator (3 corrections)

RAID page said the 10% overhead covers the decimal-TB vs binary-TiB gap, which the tool applies again separately.
How-it-works said the overhead % includes the TB/TiB gap, which the tool also applies separately.
How-it-works said RAID 1 mirrors drives in pairs.

Wedding & Party Alcohol Calculator (2 corrections)

Wedding alcohol page said a 750 ml spirit bottle gives about sixteen measures.
Wedding alcohol page said the model assumes roughly one drink per person per hour.

Flesch Reading Ease Calculator (1 correction)

Worked example said 12 words, ~1.05 syllables/word, score 98.7.

Flesch–Kincaid Grade Level Calculator (1 correction)

Worked example said grade 17.4.

Gunning Fog Index Calculator (3 corrections)

Worked example said 11 words, 4 complex, score 18.4.
FAQ said the tool excludes proper nouns and closed compounds like "everywhere".
How-it-works said the tool makes allowances for proper nouns and compound words in general.

SMOG Index Calculator (1 correction)

Worked example said 14.2 and listed six words (including 2-syllable "symptoms") as five polysyllables.

Coleman–Liau Index Calculator (2 corrections)

Worked example said 8.5, ~4.6 letters/word, ~7 words/sentence.
Said Coleman-Liau uses characters per word.

Readability Consensus Score (2 corrections)

Worked example said eight formulas.
Said the consensus combines six measurements.

Raygor Readability Graph Calculator (1 correction)

The Raygor FAQ said the graph plots letters per 100 words.

Automated Readability Index (ARI) Calculator (2 corrections)

The ARI page's worked example gave 9.6.
The ARI example said 4.4 characters per word and a grade near 10.

Linsear Write Readability Calculator (3 corrections)

The Linsear Write page's example had 70 easy and 12 hard words in a '100-word' sample and gave grade 8.5.
The example output was grade 8.5.
The example said the total is divided by sentences then halved to 8.5, omitting the −2 step.

LIX Readability Calculator (2 corrections)

The LIX page's Swedish example gave 39 (newspaper prose).
The example said the sentence sits in the 30s–40s news range.

Syllable Counter & Analyzer (2 corrections)

The syllable counter page's example said 17 syllables (1.89 per word).
The example said photosynthesis was nearly a third of the total.

Cloze Test Generator (2 corrections)

The cloze page's example blanked 'sat' and 'the' for every 3rd word.
The cloze page said the tool automatically leaves the first and last sentences intact; it blanks across the whole passage.

Spache Readability Score (1 correction)

The Spache page's example gave grade 1.4.

McAlpine EFLAW Readability (6 corrections)

The EFLAW page's example gave 17.0 (very easy), and its explanation contradicted itself with 23.
The EFLAW example counted eight mini-words, got 23 and said 'Wait' mid-explanation.
The EFLAW page said the formula rewards short common words; it penalises them.
The EFLAW page said the formula penalises long words by length; it ignores them.
The EFLAW comparison said it rewards short common words.
The EFLAW how-it-works said it rewards short common words as the easiest for ESL readers; the formula penalises them as hard.

Powers-Sumner-Kearl Readability (1 correction)

The PSK page's example (8 words per sentence, 110 syllables per 100 words) gave grade 2.8.

Fry Readability Graph Calculator (2 corrections)

The Fry page said the tool reports the graph region the point lands in and that 'the arithmetic is the same'; the tool uses one 100-word sample and a linear approximation.
The Fry page implied the tool reports Fry's 'invalid' region; it never does.

Lexical Density Calculator (1 correction)

The lexical density worked example said 7 of 11 words are content words, 64%.

Adverb / -ly Word Detector (4 corrections)

The adverb worked example said 2 adverbs in 9 words (22%).
The adverb tool page said "apply", "only" and "likely" are excluded and that strict mode requires a recognisable adjective stem; the tool flags all three words and strict mode only drops stems shorter than three letters.
The adverb FAQ said strict mode gives tighter detection "at the risk of missing neologisms", implying a stem dictionary; it only drops very short -ly words.
The adverb "how it works" said strict mode checks for an adjective stem; it only skips -ly words with a stem under three letters.

Cliché Detector (3 corrections)

The cliché worked example labelled "at the end of the day" as weasel.
The cliché tool page gave "many experts believe", "studies show" and "it is widely accepted" as the weasel phrases the tool catches; none is on its list.
The cliché tool page used "when all is said and done at the end of things" as a variant the tool misses; it contains "when all is said and done", which is on the list.

Filler / Weak Word Detector (2 corrections)

The filler worked example said 5 fillers in 16 words (31%).
The filler tool page gave "of course" as a discourse marker the tool flags; it is not on the list.

Letter Frequency Analyzer (3 corrections)

The letter frequency worked example showed E expected at 12.7%; the tool's baseline is 12.49%.
The letter frequency FAQ gave the English baseline for E as ~12.7%; the tool's baseline is 12.49%.
The letter frequency "how it works" gave E's expected frequency as ~12.7%; the tool uses 12.49%.

Headline Analyzer (3 corrections)

The headline worked example said 88/100 with 1 power and 1 emotional word ("boost") and "good length (40 chars)".
The headline FAQ gave "amazing", "terrifying" and "beloved" as emotional words; the tool tags them power, uncommon and uncommon.
The headline "how it works" said the score uses five axes including readability of the headline and word count; neither is scored.

Text Entropy & Predictability Analyzer (2 corrections)

The entropy worked example said 2.32 bits, 6 unique words, "Very predictable (predictability 51%)".
The entropy FAQ said typical prose gives ~9–10 bits per word; on the text lengths people paste, the tool gives about 8 and cannot exceed log₂(unique words).

Keyword Density Analyzer (1 correction)

The keyword density page said excluding stopwords "fragments" multi-word phrases like "return on investment"; the tool keeps any phrase with at least one content word.

Tweet / X Length Checker (3 corrections)

FAQ said Arabic/Hebrew characters count as 2.
How-it-works said Arabic/Hebrew code points count 2.
Worked example claimed 67 weighted characters (213 remaining).

SEO Meta Tag Length Checker (3 corrections)

Worked example quoted the title as 47 chars / 458px (it is 44 chars / 367px) and a 545px description as 'short', which the tool would rate 'good' (short is under 480px).
How-it-works said E is a narrow character.
Tool page body said the desktop description budget is about 920px while the tool itself flags 'long' only above 960px.

Sentence Length & Variety Analyzer (2 corrections)

Worked example claimed sentence lengths 2/17/2, mean 8.7 and SD 6.3.
Tool page said the tool outputs a list of longest sentences.

Word Frequency Analyzer (1 correction)

How-it-works said the CSV export contains the frequency table.

Lexical Diversity (TTR) Calculator (1 correction)

Worked example claimed TTR 0.45 from 5 unique of 11 words, labelled Low diversity.

Hedge Word Detector (1 correction)

Worked example explanation named five hedges including 'possible', which the tool does not count (it reports 4).

Sentence Complexity Analyzer (3 corrections)

Worked example claimed average clause depth 1.67.
Pitfall said long sentences without subordination are never flagged; the tool flags any sentence over 35 words.
FAQ said split suggestions pick sentences that combine passive voice with deep subordination.

Text Formality Analyzer (3 corrections)

Worked example claimed a score of 14 (very casual) with heavy contractions and multiple first-person pronouns.
FAQ listed sentence length as a formality input; the tool uses average word length, not sentence length.
Tool page said the tool detects phrasal verbs, nominalisation and longer noun phrases.

Vocabulary Sophistication Analyzer (2 corrections)

Worked example claimed 78% advanced, seven of nine words, average word length over 9.
Tool page said vocabulary is scored by general English usage-frequency bands.

Sentence & Paragraph Counter (2 corrections)

Tool page said the tool reports words per paragraph and its distribution.
FAQ said a sentence break needs a following capital letter and that decimals confuse the detector.

Text Cohesion & Transition Analyzer (2 corrections)

Worked example quoted 'Cohesion: 28 / 100' for a 17-word input the tool refuses to score (30-word minimum); the formula gives 30 for it anyway.
Page said the measure looks at repeated and related vocabulary between adjacent sentences; the tool does not measure that.

AI Text Pattern Detector (2 corrections)

Example claimed 72 / 100 'Many AI-style patterns' and 'heavy hedging'; the tool scores this input 53 ('Some patterns') with a hedge rate of 0.
Page said the tool detects evaluative adjectives, tidy triads and summary sentences; it measures none of those.

Text Consistency Analyzer (5 corrections)

Example claimed '2 issues — UK/US spelling mix; mixed date format' for a 21-word text the tool refuses (30-word minimum) and that would yield 1 issue (one date can't be a mixed format).
Pitfall said ~35 spelling pairs; the list has 29.
FAQ said about 35 pairs; the list has 29.
How-it-works said ~35 pairs; the list has 29.
Page said the tool checks serial commas and words-vs-digits number formatting; it checks neither (and the page's own pitfall says no Oxford-comma check).

Pictionary Word Generator (1 correction)

Page referred to Pictionary 'difficulty tiers'; the tool has categories, not difficulty tiers.

Comic Script Page-to-Panel Pacing Calculator (1 correction)

Comic pacing worked example said 6 panels / 40 words / 6 beats rates dialogue 'balanced' and action 'at the edge'

Book Length & Writing Goal Planner (1 correction)

Book planner explainer said the projection assumes you write every day

Fantasy Calendar Generator & Date Converter (4 corrections)

Fantasy calendar worked example said Month 4, day 22, weekday 6
Fantasy calendar explainer said the default calendar includes 142 intercalary days
Fantasy calendar explainer discussed 'the intercalary days' as if the tool had them
Fantasy calendar FAQ said dates are added across 'your custom month and leap-year rules'

Script Read Time / Voiceover Timer (1 correction)

Script read-time explainer said the tool defaults to a custom pace of 150 wpm

Cosplay Fabric & Material Estimator (2 corrections)

Cosplay worked example used a 170 cm radius and ~9 m² → '~6.5 m, round to 7 m', contradicting its own '~5 meters' output
Cosplay how-it-works said the full-circle cape is π × radius²

Screenplay Screen-Time Estimator (3 corrections)

Screenplay worked example said 90 action pages at 50/40/10 run 1:48:00 (108 minutes)
Screenplay explainer said the Drama preset is 1.0 across the board
Screenplay FAQ said dialogue is always roughly 1:1 and the genre applies a final multiplier

Conlang Syllable & Phonotactics Checker (5 corrections)

Conlang worked example claimed aman, katira, mendor, xthun score 3 of 4; aman has no onset and mendor uses an undeclared 'd', so only katira fits
Conlang pitfalls described greedy matching
Conlang how-it-works described greedy syllable splitting
Conlang FAQ said units are matched greedily by longest match
Conlang FAQ said the matcher prefers the longest unit

Fantasy Currency & Economy Balancer (3 corrections)

Currency how-it-works said grimdark multiplies prices by ×0.5
Currency worked example said grimdark roughly doubles everything except wages
Currency explainer said the tool converts either a tidy or a messy (12/20-style) ratio

Fantasy Bloodline & Trait Calculator (1 correction)

Genetics worked example labelled the carrier cross 'AA × AA'

Dice Bag Capacity & Weight Estimator (1 correction)

Dice bag worked example put 15 acrylic sets in 'Dragon hoard' territory

TCG Booster Pack Drop-Rate Calculator (1 correction)

TCG worked example said ~88 packs for 95%

3D Printer Electricity Cost Calculator (2 corrections)

FAQ said 2.4 kWh/month of standby costs "a few dollars" a month.
FAQ said resin and FDM energy per print can be similar once curing is included.

Soap Making Lye & Cost Calculator (2 corrections)

Soap lye worked example said 108.9 g lye, $13.75 batch, $1.38/bar (the lye figure does not follow from its own formula).
Soap lye example explanation said the formula gives ≈108.9 g and a $13.75 batch.

Guides corrected

Each now shows an "Updated" date under its title that links back here.

What does "28mm scale" actually divide by? The hidden ratio behind heroic-scale terrain (1 correction)

Described the extra foam layer on an exact fit as current behaviour; the tool was fixed the same day.

Why you can’t talk to Mars in real time: a sci-fi writer’s guide to light-speed lag (1 correction)

Said Venus's one-way delay reaches over fifteen minutes; it peaks at about fourteen.

How long does it take to cross a kingdom? A travel-time guide for writers and GMs (1 correction)

Gave terrain slowdowns and sailing distances well beyond the calculator's presets (×0.75 and ×0.5, and 40 miles a day by ship at 8 hours) without saying how to get them.

Dice pool probability for tabletop games: why the average roll lies to you (1 correction)

Said the calculator models exploding dice; it does not.

Pick the odds first: setting dice target numbers backward from the success rate you want (1 correction)

Said the calculator handles exploding pools; it does not.

How much paint you actually need for an army (and why "eyeballing it" runs out mid-batch) (1 correction)

Said a drybrush pass on a terrain piece uses less paint than one infantry basecoat; by the tool's own multipliers it uses 2.4 times as much. Its heading said a ten-model squad while the example used twenty.

How many blacksmiths does a fantasy town have? Building believable settlements (1 correction)

Said a metropolis absorbs a 100,000 gp sale; the calculator's limit is 40,000 gp (80,000 when wealthy). It also said a hamlet cannot support a blacksmith without explaining why the calculator shows one.

Is that craft fair booth actually worth it? Do the math first (1 correction)

The craft fair post called 67 cents a mile "a standard mileage rate"; that was the 2024 IRS rate.

How to set your freelance rate (the math most freelancers skip) (2 corrections)

The freelance rate post told readers to multiply the hourly rate by fewer than 8 'realistic' billable hours per day, which discounts unbillable time a second time (e.g. $575.76/day at 6 hours vs the $767.68 the year needs).
Weeks-off example: two weeks vacation plus three slow weeks gave "47 nominal weeks and 44 real ones".

Why herringbone tile needs more material than you think (3 corrections)

The herringbone waste post said a straight lay needs about 199 tiles; 180 x 1.10 is 198.
The herringbone post claimed every tile in the field is cut at an angle, which contradicts the tool page's FAQ: only border tiles are cut.
The post described herringbone corner cuts as 90-degree cuts.

French drain materials: gravel, pipe, and fabric by the numbers (1 correction)

The French drain materials post said an 18-24 inch trench gets below frost heave in cold climates.

Epoxy garage floor coating: how many kits do you actually need? (1 correction)

The epoxy kits post described a kit's 250 sq ft rating as its total output across all coats, then multiplied by coats anyway, contradicting the calculator, which treats it as coverage per coat.

How a "simple" Game Boy homebrew blows through its cartridge in a week (2 corrections)

Retro ROM post said NES and Genesis games routinely shipped multiple megabytes through bank switching.
Retro ROM post promised bank-switching and VRAM warnings.

How much disk will your game audio eat? The PCM math, explained (1 correction)

Game audio post said a 192 kbps MP3 is about 8% of the WAV size.

RAID capacity explained: why your NAS shows less space than the drives promise (1 correction)

RAID post said an n-drive RAID 1 yields floor(n/2) drives of space.

Flesch vs Gunning Fog vs SMOG: which readability formula should you use? (1 correction)

Said 206.835 is "the score of pure first-grade vocabulary" and 0 is the bottom of the scale.

How to lower your readability score: a practical editing playbook (2 corrections)

Said "The implementation of the policy" is six words.
Called articles "zero-syllable noise" in the formulas.

Every readability formula explained, briefly (1 correction)

Said the consensus calculator applies every formula in the post (incl. LIX, Fry, Raygor) and averages them.

How reading time is calculated (and why every blog uses a different number) (2 corrections)

Said Medium adds 11 seconds for every image after the first.
Said German is read faster per word than English.

The Gunning Fog Index, explained: what the score means and how to lower it (1 correction)

Said the Fog calculator shows your complex-word percentage.

The SMOG index: the readability formula health writers actually trust (1 correction)

Gave "processes" as a word SMOG counts but Fog skips; our syllable counter scores it as 2 syllables, so SMOG does not count it either.

Dale–Chall readability explained: the formula that scores words you actually know (3 corrections)

The Dale–Chall guide said 'knowledge' is on the familiar list and scores as easy; the calculator flags 'knowledge' as difficult.
The Dale–Chall guide said 'use' and 'utilise' score nearly the same on Flesch, but 'utilise' has three syllables and Flesch scores it as harder.
The guide suggested swapping 'epistemology' for 'knowledge' to clear a flag, but the calculator flags 'knowledge' too.

The Automated Readability Index: why counting characters beats counting syllables (1 correction)

The ARI guide's worked example gave 4.4 characters per word and 14 words per sentence as grade 9.6; those inputs actually give 6.3.

Linsear Write: the readability formula that counts easy and hard words by hand (2 corrections)

The Linsear Write guide said the calculator accumulates words until it reaches 100; it actually stops at the last whole sentence that fits within 100 words.
The Linsear Write guide said the 20-point seam sits well above ordinary prose; it sits at about grade 9, the top of the general-audience range.

The Spache Readability Formula: measuring text for the youngest readers (1 correction)

The Spache guide said the familiar list has roughly 750 words.

Powers-Sumner-Kearl: the readability formula built for beginning readers (1 correction)

The PSK guide said 'iguana' is two syllables; it is three, and the tool counts three.

How to read the Fry Readability Graph (and why it still beats a single grade number) (1 correction)

The Fry guide said the calculator gives the same answer as reading the printed graph; its grade comes from a linear approximation that can differ by a couple of grades.

What is a good keyword density? The honest answer for modern SEO (1 correction)

The keyword density guide said the bigram "how to" survives the stopword filter; the tool drops it because both words are stopwords.

How a headline analyzer scores your title (and what to do with the number) (4 corrections)

The headline guide description said the score is "five measurable things added together".
The headline guide said the score has five axes, including "the headline's own readability" and word count; the tool computes no readability and the word count does not affect the score.
The headline guide gave "amazing" and "astonishing" as emotional words; the tool tags "amazing" as a power word and "astonishing" as uncommon.
The headline guide referred to "the five axes".

How to find and fix the clichés hiding in your writing (4 corrections)

The cliché guide gave "the tip of the iceberg" and "the bottom line" as idioms the tool catches; the first is filed as a metaphor and the second is not on the list.
The cliché guide listed "a perfect storm", "the elephant in the room" and "level the playing field" as dead metaphors the tool hunts; none of them is matched by the tool.
The cliché guide listed "studies show" as a weasel phrase the tool catches; it is not on the list.
The cliché guide listed "move the goalposts" as a sports cliché the tool catches; it is not on the list.

The -ly trap: how to find and fix adverb overload (2 corrections)

The adverb guide listed "rely" among the non-adverbs the detector subtracts; the tool flags "rely".
The adverb guide said strict mode checks for a recognisable adjective stem and "cuts false positives sharply"; it only drops -ly words with a stem under three letters (e.g. "only", "rely").

Very, really, just, basically: how to find and cut filler words (1 correction)

The filler guide said dialogue is "exempt by design"; the tool flags filler inside quoted dialogue like anywhere else.

Why your 280-character tweet is actually too long: weighted counting explained (2 corrections)

The post said Arabic and Hebrew count as 2 characters on X.
The post said the checker's count always equals X's, right after an x.com example; the checker only collapses links that begin with http:// or https://, so a bare x.com counted 5, not 23.

Word frequency analysis: seeing what your text actually overuses (1 correction)

Post said the word frequency table exports to CSV; the Export CSV button only writes six summary metrics.

Measuring vocabulary sophistication: rare words, CEFR levels, and when "advanced" hurts you (1 correction)

Post called the everyday list a 1948 word list.

Why your writing feels choppy: measuring cohesion and transitions (1 correction)

Post said the score weights paragraph linkage above pronoun use; the code weights them equally (30 points each, transition density 40).

What AI writing detectors actually measure (and why they get it wrong) (1 correction)

Post said the technical preset is more tolerant of repeated terminology; the technical preset weights 3-gram repetition 25 and vocabulary repetition 20 (vs 15/15 in general), so repetition counts for more.

Why charades goes stale after ten rounds — and how a real word pool fixes it (1 correction)

Worked example had the generator draw 'assembling IKEA furniture', which is not in the charades pool.

Scattergories over a long game night: repeated categories, sticky settings, and a clock that can run slow (3 corrections)

Post cited 'Things that are sticky' as a hard-tagged category (it is not in the bank) and called U nearly dead for Countries (7 countries start with U).
Post said the boxed die drops J, Q, U, V, X, Z; the standard die drops Q, U, V, X, Y, Z (J is on it).
House-rule advice said reroll on J, U, V, based on the wrong die letters.

Is your comic page too cramped? Reading panel, dialogue, and action density (1 correction)

Comic pacing guide called an 8-panel US page 'near the crowded end' of the range

Why "a few weeks pass" breaks your campaign timeline — and the date math to fix it (1 correction)

Fantasy calendar guide's worked example 'corrected' month 4 day 23 to day 22 and gave weekday 6

How much fabric does a cosplay actually need? Stop guessing at the cutting table (1 correction)

Cosplay fabric guide said a circle skirt's radius is the waist circumference plus the length

One page, one minute? Why your screenplay's real runtime depends on what's on the page (3 corrections)

Screenplay guide called 110 pages → ~125 minutes a 'twelve-minute overshoot'
Screenplay guide said the sitcom multipliers explain a 32-page script filling 22 minutes, but those multipliers give about 17 minutes for 32 pages
Screenplay guide said the tool flags a 75-page script that is half set pieces; it only flags action above 50%

Why your invented language's names don't sound like they belong together (1 correction)

Conlang guide's example said 'aman' and 'mendor' pass a CV/CVC p-t-k-m-n-s-r-l ruleset; 'aman' has no onset consonant and 'mendor' uses 'd', so both fail

Building a fantasy economy that actually holds together (1 correction)

Currency guide said 1:10:100 mirrors pounds, shillings and pence

Punnett squares for fictional bloodlines: a genetics primer for writers (1 correction)

Genetics guide said the simple model fails to predict two brown-eyed parents having a blue-eyed child, though that is exactly the carrier cross it predicts

The real odds of pulling that chase card (and why the box let you down) (1 correction)

TCG odds guide said the calculator takes a set size, pack composition and rarity ratios

KDP royalty tiers explained: why some ebooks earn 70% and others only 35% (14 corrections)

KDP tiers guide opened with a $4.99 ebook earning $3.04 and a $12.99 one earning $4.55 at 35%.
KDP tiers guide gave the 70% band as roughly $2.99–$9.99.
Said a $12.99 ebook drops to 35%.
Used $12.99 as the out-of-band price.
Described the $0.15/MB fee as subtracted from the royalty.
Said an 8 MB file loses $1.20 per sale.
Worked example: $4.99 × 70% − $0.45 = $3.04.
Said $12.99 is above the 70% band.
$12.99 × 35% = $4.55.
Thriller example at $12.99.
Said paperback royalty is a flat 60% applied after subtracting a print cost of $0.85 + $0.012 per page over 108.
200-page $12.99 paperback: print $1.95, royalty $6.62.
Minimum viable price $1.95 ÷ 0.60 = $3.26, and said the storefront lets you price below it.
Said the calculator applies the correct royalty tier automatically.

Why a cheap KDP paperback can earn you nothing: the print-cost floor explained (15 corrections)

Said paperback royalty is a share of what is left after printing.
Formula given as 60% × (list price − printing cost).
Said the 60% is applied after subtracting printing.
Print cost given as $0.85 + $0.012 per page beyond 108.
108 pages printed for $0.85.
300 pages printed for about $3.15.
500 pages printed for past $5.50.
Said printing comes out before the 60% split.
Said royalty is zero when price equals printing cost, while computing the floor as cost ÷ 0.60.
300-page example: floor $5.25, $7.99 earns $2.90, $12.99 earns $5.90.
Said trimming pages widens the gap the 60% applies to.
500-page book: $5.50 print cost, floor near $9.17.
Said a $14.99 paperback keeps ~$6.
Referred to "$0.85-plus-per-page figures".
Summarised the mechanism as "60% of whatever's left".

Exclusive vs. non-exclusive ACX distribution: the real royalty math (15 corrections)

Intro: exclusive self-narrated earns $79.80/mo, ~$958/yr (40%).
Intro: royalty-share author nets $39.90/mo.
Exclusive rate given as 40%.
Non-exclusive rate given as 25%.
"The 40% rate looks like the obvious winner".
Said the narrator takes a cut for the life of the audiobook and the 50/50 split is negotiable.
Said royalty share halves the take forever and never reverts.
"standing obligation on every future sale".
Scenario 1: $7.98/sale, $79.80/mo, $957.60/yr.
Scenario 2: $3.99 each, $39.90/mo, $478.80/yr.
Scenario 3: 25%, $4.9875/sale, $49.875/mo, $598.50/yr.
Comparison: $49.88 vs $39.90/mo, 40% vs 25%.
"exclusive at the full 40%".
Called royalty share a permanent cut.
Worked the example at legacy rates on the retail price with no date.

What it actually costs to charge an EV vs. fill up with gas (1 correction)

Said moving from $0.16 to $0.40/kWh "roughly triples" the trip cost to about $38.

Why your EV draws more kWh than its battery holds — charging losses explained (2 corrections)

Said 10–15% losses on a 75 kWh pack means 84–86 kWh from the wall.
Said lower efficiency raises cost per mile "twice over" because losses are computed on the energy drawn.

What your 3D printer actually costs in electricity (including standby) (1 correction)

Said 24-hour 12W standby "adds 8.6 kWh on top of 16 hours" and is about a third more than a printer powered down between jobs.

The real cost of a handmade candle (it's not just wax) (2 corrections)

Candle cost guide said wax, fragrance and hard goods were "barely over half" the $9.33 unit cost.
Candle cost guide called a $12 price on a $9.33 cost "a 29% margin"; 29% is the markup over cost.

How to price a 3D print: the costs every maker forgets (2 corrections)

3D print pricing guide said the calculator returns a wear allowance and a tiered retail price
3D print pricing guide said filament is the smallest cost of a print, while its own example puts electricity at 19 cents against $1.50 of filament

How to price art commissions without guessing (1 correction)

Commission guide said the commercial license alone accounts for almost $1,000 of the price difference

Your laser-cut wholesale price might be below cost — here is the markup that fixes it (1 correction)

Laser wholesale guide quoted the 50%-markup wholesale price as $11.30

October 12, 2026: guides on a regular schedule, and this page

New guides now come out twice a week, on Tuesdays and Fridays, instead of in batches. Each one follows a whole job that needs several calculators in a row, such as taking a novel from word count to ebook, paperback and audiobook prices, or deciding whether a craft fair booth beats selling the same candles online. Single-tool explainers already exist for all 104 tools. Every number in the new guides was produced by running the calculators themselves, and writing them is how the corrections above were found.

Food and drink for a 120-guest evening wedding, planned as one shopping list (October 20, 2026)
Pricing articulated dragons from a two-printer farm: power per print, unit cost, then Etsy fees (October 16, 2026)
Prepping a D&D travel session: road days, fight length and check odds (October 13, 2026)
The pre-publish check for a how-to blog post: readability, keywords, meta tags and reading time (October 12, 2026)
Craft fair booth or Etsy listing? Costing one soy candle and running both sales channels (October 12, 2026)
One 90,000-word novel, three formats: from draft plan to Kindle, paperback and audiobook royalties (October 12, 2026)

The sitemap now reports the date each page last changed rather than the date the site was last rebuilt, the blog feed lists only published guides, and every tool page links here.

September 7, 2026: ads off pages with nothing to read, and a byline on every guide

Advertising code was removed from pages with no content of their own (the 404 page, Contact, Privacy, Terms, and blog listing pages too short to justify it). Guides now name their author, and the home page stopped repeating the tool directory word for word: it shows one card per category instead.

August 27, 2026: every tool page explains its limits

Each of the 104 tool pages gained a longer explanation written from the tool's own code: what it assumes, where its model stops being accurate, a worked example, and the mistakes people make most often. Some of these say where we depart from the published method on purpose. The SMOG page, for instance, explains that this implementation scales polysyllables over the whole text instead of taking the formula's 30-sentence sample, so short passages are unreliable.

August 26, 2026: the calculators are in the page, and sources are cited

Until this date the 55 calculators were built entirely in the browser, so anyone without JavaScript (and any search engine that did not run it) saw a page with no calculator on it. Each calculator is now rendered into the page itself with its default inputs and result. The same release added sources: the 16 readability formulas cite the published work they implement (Flesch 1948, the Kincaid Navy report of 1975, McLaughlin's SMOG of 1969 and the rest), and the selling calculators link the fee pages of Etsy, Amazon KDP and ACX and say plainly that those rates change. Each tool page shows the date its formula or rates were last verified.

August 14, 2026: English only, and tools reorganised

Machine-translated copies of every page in seventeen languages were removed rather than left online. They had never been reviewed by a person, and 2,232 of them were English guides served under another language's label. The 104 tools were regrouped into nine categories that match what they do, and every category is now listed on the home page and the tool directory; 47 maker, creator and game-development tools had not been reachable by browsing before. Tool pages gained their own meta descriptions and links to their guides, and pages load about half as much script.

July 5, 2026: 27 maker, home and selling calculators

The selling, home-improvement and home-tech calculators were added, from the Etsy fee calculator to the NAS RAID capacity calculator, each with a guide.

June 3, 2026: thirty-five new tools

Thirty-five tools were added alongside the original readability formulas: seven more text-analysis tools and calculators for writers, game masters, game developers and makers, from fantasy travel time to 3D print pricing. The first guides and the legal pages came with them.

All guides by month

Every tool has at least one guide. These are all 130 published guides, newest first.

October 2026: 6 guides

August 2026: 2 guides

July 2026: 77 guides

June 2026: 34 guides

May 2026: 8 guides

April 2026: 3 guides

How corrections are handled

A wrong answer is fixed in the tool and in its explanation at the same time, and it is recorded here with the old and new numbers. We don't quietly edit the past. If a guide described behaviour that has since changed, the guide is corrected and shows an "Updated" date under its title with a link back to its entry on this page. More on how the tools are checked is on the About page.