Mental Calculation Techniques of World Champions: Left to Right, Criss-Cross, Squares and Calendars

World champion mental calculators do not use secret methods. They use a small set of ordinary techniques, worked from the left so the answer comes out in the order it is spoken: left to right addition, left to right and criss-cross multiplication, squaring by moving to a round number, digit patterns for roots, and a memorised calendar rule. What makes them fast is drill on those few techniques, and the measured results show how fast that is: at the 2024 Mental Calculation World Cup the winner named the weekday of 100 random dates in one minute and solved 25 net multiplications of two 8-digit numbers in ten minutes.

This post works every technique on the fixed problem set used across our mathematics series, with every intermediate number checked in Python, teaches John Conway's Doomsday rule for the day of the week, and reports only records measured under competition rules, with sources.

Key takeaways

• The Mental Calculation World Cup has run since 2004. Its core events: ten 10-digit additions, 8 x 8 digit products, 6-digit square roots and calendar dates.
• Champions work from the left: the leading digits matter most, can be written at once and give an early size check.
• Criss-cross turns a 3 x 3 digit product into five column sums and a running total.
• Squaring near a round number uses one identity: a² = (a + b)(a − b) + b². With it, 85² is 90 x 80 + 25 = 7,225 and 88² is 90 x 86 + 4 = 7,744.
• Cube roots of perfect cubes up to six digits take two lookups.
• The Doomsday rule finds any Gregorian weekday from a century anchor, three small numbers and easy month dates.
• The measured records show what years of drill on these methods can do, and nothing more.

The problem set for this series

This post belongs to the fast calculation series, which follows our mathematics traditions series. Every post in both series works the same problems, so you can compare methods directly:

The traditions posts work the same set on other systems: the Indian and Vedic mathematics post, the Chinese mathematics post with counting rods, the Japanese mathematics post with the soroban and anzan, and the three way comparison.

Every number in the tables below was produced by a short Python script that redoes each step the way the method does it, then compares the final value against ordinary integer arithmetic. Every date in the calendar section was checked against Python's datetime module.

What is the Mental Calculation World Cup?

The Mental Calculation World Cup (MCWC) is an invitational competition for people who calculate without paper aids beyond writing down the answer. Ralf Laue, who runs the recordholders.org site, organises it. The organiser's archive lists editions from 2004 to 2024, roughly every two years, and the 2024 information sheet calls that edition the tenth. Robert Fountain of Great Britain won the first one, in Annaberg-Buchholz, Germany, on 30 October 2004.

The 2024 edition ran from 13 to 15 September 2024 at the Heinz Nixdorf MuseumsForum in Paderborn, Germany, with the contest itself on 14 and 15 September. The organiser's page says there were 69 registrations from 21 countries, 40 qualified and 35 calculators from 17 countries took part. (A report on worldmentalcalculation.com gives 16 countries; the organiser's figure is used here.) The field is capped at 40. Applicants improve their chance of a place by submitting earlier competition results or results verified with the Memoriad software and confirmed by a witness.

The four core events and the surprise tasks

The 2024 information sheet names the categories: addition, multiplication, extracting roots, calendar from memory and surprise tasks. The results page states the format of each event:

The overall champion has the best combined score; separate titles go to the winner of each category and to the best performer in the surprise tasks.

The penalty for wrong answers matters: 20 right and 5 wrong scores 15. The events reward accuracy at speed, which is why checking is part of every champion's method.

Winners and results from the organiser's pages

In 2024 Kaloyan Geshev (Bulgaria) was second with 532.86 points and Hiko Chiba (Japan) third with 455.75. Shukla also took the most versatile calculator title with 419.84 points from the surprise tasks.

Records set with the Memoriad software at the 2024 World Cup

Alongside the main contest, record attempts ran on the Memoriad software, which times each attempt and checks each answer. The worldmentalcalculation.com report of the event lists three:

That report asks readers not to treat it as an official record source, so these are reported times, not ratified records. It also mentions a calendar on paper record by Shukla without a figure.

The Memoriad

The Memoriad is a separate competition that combines memory and mental calculation events and supplies the software used for many record attempts. According to the worldmentalcalculation.com results report, Memoriad 2024 was held from 6 to 10 November 2024 in the United Arab Emirates, with six calculation disciplines (additions, multiplications, square roots, division, calendar dates and flash anzan) and six memory disciplines. The calendar winner was reported as Jay Baldiya Jain with 125 dates in 51.24 seconds, and later record attempts were reported at 27.3 seconds for ten square roots (Shashank Jain) and 29.6 seconds for ten divisions (Kaloyan Geshev).

Flash anzan, the event of adding numbers flashed on a screen, comes straight from soroban training. The Japanese mathematics post explains how anzan works and how it is graded.

What the records say about pace

Turning the measured results into time per task shows what "fast" means here. These are plain divisions of the published numbers, checked in Python:

An 8 x 8 digit product by criss-cross needs 64 single-digit multiplications, so 24 seconds per product is about 2.7 a second, plus the column additions. The score is net of wrong answers, so the true time per attempt was shorter. These are the best people in the world on the day: they show what is possible, not what a learner should expect.

Famous calculation feats that were never timed under published competition rules are left out of this post on purpose. A record is only useful when you know the task, the time limit and who checked the answers.

Why do champions calculate from left to right?

School arithmetic runs right to left because carries flow leftwards, and paper makes that cheap. In your head, right to left has three problems:

1. The answer comes out backwards. You must hold every digit until the end, then read them out in reverse.
2. You learn the size of the answer last, although a slip in the leading digit is the most costly error.
3. Memory load grows with the length of the answer, because every settled digit has to be stored.

Left to right fixes all three. The leading digits come first, they can be said or written immediately, and an early estimate catches gross errors. The cost is that a later carry can change a digit you have already settled. Champions handle that in two ways: hold one digit back until the next column shows whether it carries, or keep a running total and add each new column into it.

Left to right addition, worked

Add the six answers from the problem set: 9,506 + 2,491 + 202,264 + 7,225 + 88 + 200. A mental calculator keeps one running total and adds each number into it from the left.

The total is 221,774. Notice step 2: adding 91 to 11,906 is easier as "add 100, take 9" (11,906 + 100 = 12,006; 12,006 − 9 = 11,997). Champions switch freely between adding and subtracting complements whenever a number sits near a round value.

Column addition from the left: the ten-number task

The World Cup addition task is ten numbers of ten digits each, stacked in a column. Adding them one at a time means holding a running total of up to eleven digits through ten additions. The faster approach, used by many competitors, is column by column from the left: sum each column of digits, then fold the column sum into a running total by shifting it one place.

A four number, seven digit example shows the layout:

   3 8 1 5 2 0 7
   4 9 2 6 3 8 1
   1 5 0 7 6 9 4
   8 2 6 3 9 1 5
   -------------
column sums (left to right): 16  24  9  21  20  18  17

The sum is 18,513,197, and Python agrees. Column sums stay small (at most 90 for ten numbers); only the running total grows. With ten numbers, all the columns still to come add less than ten units of the running total's last place. So every digit of the running total except the last is already final, unless a run of 9s sits just before it, and a competitor can write the answer out left to right while still calculating.

Left to right multiplication

The splitting method

The plainest champion method is the distributive law, worked from the biggest piece down. Split one factor into a round part and a small part, multiply each, and add from the left.

98 x 97. Treat 98 as 100 − 2 or treat 97 as 100 − 3. With 97 = 100 − 3:

47 x 53. Split 53 into 50 + 3:

386 x 524. Split 524 into 500 + 24, then 24 into 20 + 4:

The splitting method is easy to learn because it is just the distributive law, but the number of partial results you hold grows with the size of the factors. For two 8-digit numbers it is not practical, which is why competitors use criss-cross for big products.

Criss-cross multiplication

Criss-cross, also called cross multiplication, computes the product one column at a time. Write the two factors one above the other. The leftmost column of the answer is the product of the two leading digits. Each later column is a sum of cross products: every pair of digits whose positions add up to that column. The rightmost column is the product of the two units digits.

For two 2-digit numbers AB x CD:

  column 1:  A x C
  column 2:  A x D + B x C      (the cross)
  column 3:  B x D

Then fold the columns together from the left, multiplying the running total by 10 each time, exactly as in column addition. This is the same structure as long multiplication; the only difference is the order in which the single-digit products are added. It is the method that Bharati Krishna Tirtha's 1965 book Vedic Mathematics calls Urdhva Tiryagbhyam, "vertically and crosswise". The Indian and Vedic mathematics post explains where that book's sutras come from and why historians do not date them to the Vedas.

98 x 97 by criss-cross.

     9   8
  x  9   7
  --------
  9x9 | 9x7 + 8x9 | 8x7
   81 |  63 + 72  |  56
   81 |    135    |  56

47 x 53 by criss-cross.

     4   7
  x  5   3
  --------
  4x5 | 4x3 + 7x5 | 7x3
   20 |  12 + 35  |  21
   20 |     47    |  21

386 x 524 by criss-cross. Three-digit factors give five columns. Label the digits 3, 8, 6 on top and 5, 2, 4 below.

        3   8   6
     x  5   2   4
     ------------
col 1:  3x5                     = 15
col 2:  3x2 + 8x5               = 6 + 40        = 46
col 3:  3x4 + 8x2 + 6x5         = 12 + 16 + 30  = 58
col 4:  8x4 + 6x2               = 32 + 12       = 44
col 5:  6x4                     = 24

The answer is 202,264. After step 2 you already know the product is at least 196,000, and the three columns still to come can add at most about 26,000, so the answer is close to 200,000: that is the size check, done before most of the work. After step 4 the running total is 20,224, and column 5 can add at most 81 to 202,240, which can change only the last three places; the first three digits, 202, are final. That is the left to right advantage in one table: the leading digits arrive first and stay put.

Scaling to the 8-digit competition task

Two 8-digit factors give 15 columns. The middle column has 8 cross products, and the column sums rise from 1 product to 8 and back down to 1. The full count is 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 7 + 6 + 5 + 4 + 3 + 2 + 1 = 64 single-digit products. Each column sum is at most 8 x 81 = 648, and the running total grows to 16 digits, so competitors write each digit as soon as it is final and hold only the unsettled tail. A 24-second 8 x 8 product is 64 facts recalled instantly, 15 column sums and a running total written out as it settles.

Near a base: the shortcut that only works sometimes

98 x 97 has a second, much shorter route because both factors sit just below 100. Write each factor's distance from 100:

  98   −2
  97   −3
  ---------
  98 − 3 = 95  |  (−2) x (−3) = 06
  answer: 95 | 06 = 9,506

That is two small operations. The comparison post shows the same trick under its Vedic name, Nikhilam, and shows where it breaks: 47 x 53 is far from 100 and 386 x 524 is far from any shared base, so the distances become large numbers and the shortcut stops saving work. Champions keep it as a special case, not as their main method.

Squaring near round numbers

The identity behind it

Every squaring trick in this family comes from one identity:

a² = (a + b)(a − b) + b²

Choose b so that one of a + b or a − b is a round number. Multiplying by a round number is easy, and b² is a small square you already know. The method is taught in many mental calculation books, including Arthur Benjamin and Michael Shermer's Secrets of Mental Math (2006).

85². Take b = 5, so a + b = 90 and a − b = 80.

The familiar "squares ending in 5" rule is a special case of the same identity. For any number ending in 5, b is always 5, the two neighbours are the round tens either side, and the result is (tens digit x next digit) followed by 25: for 85, 8 x 9 = 72 followed by 25 gives 7,225.

More squares, worked the same way

The last row is the check for the square root problem below: 88² is 7,744.

47 x 53 as a difference of squares

The same identity read backwards gives (a + b)(a − b) = a² − b². 47 and 53 sit 3 either side of 50, so:

That is the fastest route to 47 x 53 in this post, and it only works because the factors are symmetric around a round number. 98 x 97 is not symmetric around a whole number (the midpoint is 97.5), so for it the near-base method is better.

Square roots

Perfect squares: √7,744

When the number is known to be a perfect square, the root comes from two lookups.

The table of last digits for squares is short. A square ends in 0, 1, 4, 5, 6 or 9, and each of 1, 4, 6 and 9 comes from two possible last digits of the root:

Step 3, comparing with the square of the number ending in 5, settles which of the two candidates is right.

Non-perfect squares to eight significant figures

The World Cup task is harder: a 6-digit number that is not a perfect square, with the root required to 8 significant figures. A standard approach is to find the nearest whole root, then refine it with a correction term. That is Newton's method written for mental use. Take √123,456 as an example (not one of the series problems, but the right size).

Python gives √123,456 = 351.3630600960, so 351.36306 is correct to 8 significant figures. Step 3 is a division of a 3-digit number by a 3-digit number to six places, and step 4 is a small refinement that only touches the last few digits. That is what the reported record of 33.78 seconds for ten such roots involves: about 3.4 seconds per root for the whole chain.

Cube roots of perfect cubes

Cube roots are a classic stage demonstration and a common surprise task. For a perfect cube up to six digits, the root has two digits and two lookups find them, because every last digit of a cube comes from exactly one last digit of the root.

The pattern is easy to keep: 0, 1, 4, 5, 6 and 9 map to themselves, and the pairs 2 and 8, and 3 and 7, swap.

You also need the cubes of 1 to 9 for the first digit: 1, 8, 27, 64, 125, 216, 343, 512, 729.

The worked examples use cubes of numbers from the problem set:

Python confirms 47³ = 103,823, 53³ = 148,877, 88³ = 681,472 and 98³ = 941,192. The method does no arithmetic beyond comparing a 3-digit number with nine stored cubes. It only works for perfect cubes; a non-perfect cube root needs the same refine-and-correct approach as the square roots above.

Division, left to right

Division is the one operation that school arithmetic already does left to right, so champions mainly speed up the inner step: estimating each quotient digit and multiplying the divisor by it.

7,429 ÷ 37

Quotient 200, remainder 29. A mental calculator compresses this into one look: 37 x 200 = 7,400, and 7,429 − 7,400 = 29.

Turning the remainder into a decimal with 37 x 27 = 999

If the decimal is wanted, a stored fact does the rest. Because 37 x 27 = 999, any fraction with denominator 37 becomes a fraction with denominator 999, and a fraction over 999 repeats its numerator in blocks of three digits.

Python confirms 7,429 ÷ 37 = 200.78378378... This is typical champion practice: a single memorised relationship (37 x 27 = 999, or 7 x 11 x 13 = 1,001) turns a long division into one multiplication.

1/19 as a decimal

The series asks for 1/19 to its full repeating period. A plain long division takes 18 steps. Champions use the fact that 19 is one less than 20, which gives two short methods.

Method 1: double from the right

Because 19 = 2 x 10 − 1, the digits of 1/19 obey a doubling rule. Start with 1 as the last digit of the period and double leftwards, carrying as you go: 1, 2, 4, 8, then 16 (write 6, carry 1), 13 (write 3, carry 1), 7, 14 (write 4, carry 1), 9, and so on. After 17 doublings the digits read 052631578947368421, and the next doubling returns to the starting 1, which marks the end of the period.

Method 2: halve from the left

The same rule run forwards is halving, which produces the digits in reading order. Write 0.0, then repeatedly divide by 2, where each new dividend is the previous remainder in front of the previous quotient digit:

The quotient digits after 0.0 are 5, 2, 6, 3, 1, 5, 7, 8, 9, 4, 7, 3, 6, 8, 4, 2, 1, which gives 0.052631578947368421 again. Each step is a halving of a number under 20, and you only ever hold one digit and one remainder.

The nines check

Because the period has an even length, its two halves add to a string of nines: 052631578 + 947368421 = 999,999,999. So once you have the first nine digits, you get the second nine by subtracting each from 9. That halves the work and checks it at the same time. The Japanese mathematics post works the same fraction by long division on the soroban, for comparison.

How do champions check their answers?

Penalties for wrong answers make checking part of the method, not an extra. Two digit checksums cost almost nothing.

Casting out nines. Replace every number by the remainder of its digit sum on division by 9. The product of the remainders must match the remainder of the product. Casting out elevens does the same with remainders on division by 11 (alternating digit sums), and catches swapped digits, which casting out nines misses.

All six checks pass. Neither checksum proves an answer right; each one only shows that an error of a certain kind did not happen. Using both catches a much wider range of slips than either alone. The Chinese mathematics post shows the remainder arithmetic behind these checks, which is the same idea as the Chinese remainder theorem.

Champions add two cheaper checks of their own:

• Size. From the left, the first two digits of a product are known early. If 386 x 524 is about 400 x 500 = 200,000, an answer of 2,022,640 is plainly wrong.
• Last digit. The last digit of a product is the last digit of the product of the last digits: 6 x 4 = 24, so 386 x 524 ends in 4.

The day of the week for any date

Calendar calculation is the most popular World Cup event with audiences, and the one with the clearest method. The usual technique is the Doomsday rule, which John Horton Conway published as "Tomorrow is the Day After Doomsday" in the Cambridge mathematics journal Eureka, issue 36, October 1973, pages 28 to 31. Earlier methods existed: Lewis Carroll published "To Find the Day of the Week for any Given Date" in Nature on 31 March 1887 (volume 35, page 517). Conway's version won out because it needs almost no arithmetic.

The idea

In any given year, a set of easy dates all fall on the same weekday. Conway called that weekday the year's Doomsday. If you know the Doomsday, any date is a short count from the nearest easy date.

Step 1: the easy dates in each month

The even months pair with themselves (4/4, 6/6, 8/8, 10/10, 12/12). The odd months pair up as 5/9, 9/5, 7/11 and 11/7. Any date in the same month as a Doomsday date is that many days away, and only the remainder on division by 7 matters.

Step 2: the century anchor

In the Gregorian calendar the Doomsday of each century year repeats every 400 years:

Numbering Sunday as 0 through Saturday as 6, the anchor for century c (1600 has c = 16) is (5 x (c mod 4) + 2) mod 7. For 1900, 19 mod 4 = 3, and 5 x 3 + 2 = 17, which is 3 mod 7: Wednesday.

Step 3: move from the century to the year

Take the last two digits of the year, y. Compute:

• a = how many 12s fit into y
• b = the remainder after removing those 12s
• c = how many 4s fit into b

Add a + b + c, take the remainder on division by 7, and count that many days forward from the century anchor. This works because a year moves the Doomsday forward one day and a leap year two, and 12 years move it forward exactly 12 + 3 = 15 days, which is one day on a 7-day cycle.

Worked dates

Each line below was checked against Python's datetime module.

14 September 2024, the first day of the 2024 World Cup contest.

20 July 1969, the first crewed Moon landing.

1 January 1600, the start of the competition range.

31 December 2100, the end of the competition range.

29 February 2000. The anchor for 2000 is Tuesday and y = 0, so 2000's Doomsday is Tuesday. The last day of February is a Doomsday date, so 29 February 2000 was a Tuesday. 2000 was a leap year because it is divisible by 400; 1900 and 2100 are not leap years because they are divisible by 100 but not by 400.

The "odd plus 11" shortcut for the year step

Chamberlain Fong and Michael K. Walters published a variant in 2010 (arXiv 1006.3913) that replaces the division by 12 with halving:

1. Take y. If it is odd, add 11.
2. Halve it.
3. If the result is odd, add 11.
4. Take the remainder on division by 7, and subtract it from 7 (a result of 7 counts as 0).
5. Count that many days forward from the century anchor.

For 1969: y = 69 is odd, so 69 + 11 = 80; half is 40; 40 is even; 40 mod 7 = 5; 7 − 5 = 2. Wednesday + 2 = Friday, the same Doomsday as before. The published results do not say which year method each competitor used.

What competitors do differently

At 0.6 seconds per date or less, nobody works through four steps consciously. After drill, the century anchors, year offsets and month dates become instant lookups, and the only live arithmetic left is a sum of small numbers mod 7. The 2004 calendar winner scored 33 correct across two one-minute series; the 2024 winner scored 100 in a single minute. The method did not change in those twenty years. The drill did.

The Gregorian caveat

The rule as given is for the Gregorian calendar, which began when Thursday 4 October 1582 was followed by Friday 15 October 1582 in the first countries to adopt it. The method agrees: anchor Wednesday for 1500; y = 82 gives a = 6, b = 10, c = 2, sum 18, which is 4 mod 7, so the Doomsday is Sunday; 15 October is 5 days after Sunday 10 October, a Friday. Countries adopted the reform at different times (Great Britain and its colonies in 1752), and Julian dates need a different anchor table. Competitions use Gregorian dates from 1600 onward.

How many steps, and how much do you hold in your head?

"Fast" should be counted, not asserted. The table below counts, for each problem, the single-digit multiplications and the number of separate partial results the method needs, and the most digits held in memory at one time when worked mentally. Counts are for the method as laid out in this post.

Digits are counted without the trailing zeros that only mark place value, so 193,000 counts as the three digits 193.

Two patterns stand out. Criss-cross does the same single-digit products as school long multiplication; it saves memory, not multiplications, because it never holds whole rows. The special-case methods cut the work to one or two operations, but only for numbers with the right shape.

What does research say about champion calculators?

There is very little peer-reviewed research on competition mental calculators, and most of it studies one or a few people. One frequently cited study is Pesenti and colleagues, "Mental calculation in a prodigy is sustained by right prefrontal and medial temporal areas", Nature Neuroscience 4(1), 103 to 107 (2001). The authors compared one expert calculator with a group of non-experts using PET imaging. Their abstract reports that the expert's skill was not simply more of what non-experts do, and that the expert could switch between effortful short-term storage and efficient long-term memory encoding and retrieval.

The limits are large: it is one expert, it describes how that person worked rather than what any training causes, and it says nothing about effects of learning these techniques, so this post claims none. It fits the methods above in one modest way: experts lean on a large store of memorised facts, which leaves little live arithmetic.

Champion techniques compared with the traditions

Champion techniques are mostly not new: criss-cross is long multiplication reordered, and the ends-in-5 square and near-base product appear in the Vedic book under other names. What is distinct is the left to right discipline, heavy use of memorised facts, picking the shortest route per problem, and timed, scored drill. The comparison post has the full method by method table for the traditions series.

Where champion techniques break down

• Large products without drill. Criss-cross for 8 x 8 digits is 64 products and 15 column sums in the head; without years of practice, paper is faster and safer.
• Special-case tricks on the wrong numbers. Near-base and difference-of-squares methods get slower, not faster, when the numbers are far from a round value.
• Roots of non-perfect powers. The lookup tricks only work for perfect squares and cubes. Everything else needs a refine-and-correct method that is real division.
• Dates outside the Gregorian calendar. The Doomsday rule needs Julian anchors for historical dates in countries that had not switched.
• Anything a computer should do. For many numbers, a calculator is faster and more reliable. Mental calculation is a skill and a sport, not a replacement for tools.

A practice plan

This plan follows the order the methods build on each other. It is a suggestion, not a measured programme; no study has tested it.

Score yourself as the World Cup does: correct minus wrong.

Common mistakes

• Dropping a carry in the running total. In left to right work a late column can carry into a digit you thought was final. Do not write a digit until the next column cannot change it.
• Using the near-base trick far from the base. For 47 x 53 near 100 the distances are 53 and 47, and the "shortcut" is harder than the original problem.
• Forgetting the zero in the right part. With base 100, the right part of 98 x 97 is 06, not 6. Writing 956 is the classic error.
• Picking the wrong root candidate. For √7,744 both 82 and 88 end in the right digit. Compare with the square of the number ending in 5 to choose.
• The wrong January Doomsday in a leap year. January's Doomsday is the 4th in leap years, not the 3rd. Century years are leap years only when divisible by 400.
• Trusting one checksum. Casting out nines misses swapped digits; add casting out elevens or a size check.

FAQ

What techniques do mental calculation champions use?

Mostly standard ones, drilled hard: left to right addition, criss-cross multiplication done left to right, squaring with (a + b)(a − b) + b², digit patterns for perfect square and cube roots, short division with memorised facts, and the Doomsday rule for calendar dates. They also memorise large tables of squares, cubes and products so that much of the work becomes recall.

What is the Mental Calculation World Cup?

An invitational competition first held in Annaberg-Buchholz, Germany, in 2004 and organised by Ralf Laue. Its core events are adding ten 10-digit numbers, multiplying two 8-digit numbers, square roots of 6-digit numbers and calendar dates, plus surprise tasks. The 2024 edition in Paderborn was the tenth.

Who won the 2024 Mental Calculation World Cup?

Aaryan Nitin Shukla of India, with 819.84 points, ahead of Kaloyan Geshev of Bulgaria (532.86) and Hiko Chiba of Japan (455.75), according to the organiser's results page.

How many dates can a champion calculate in a minute?

At the 2024 World Cup the calendar winner scored 100 correct dates in a one-minute round, with dates from 1600 to 2100. At Memoriad 2024 the reported winner did 125 dates in 51.24 seconds. In 2004 the winning score was 33 across two one-minute series.

How does the Doomsday rule work?

Each year has a weekday, its Doomsday, shared by easy dates such as 4/4, 6/6, 8/8, 10/10, 12/12, 5/9, 9/5, 7/11, 11/7 and the last day of February. Find it from the century anchor plus a + b + c from the last two digits of the year, then count from the nearest easy date to your date.

Is criss-cross multiplication the same as Vedic multiplication?

It has the structure of the method Bharati Krishna Tirtha's 1965 book calls Urdhva Tiryagbhyam, which is also the structure of ordinary long multiplication, with the column sums added from the left.

How do you find a cube root in your head?

For a perfect cube up to six digits, the cube's last digit gives the root's last digit (2 and 8 swap, 3 and 7 swap), and the thousands part gives the first digit. 103,823 ends in 3 and 103 lies between 64 and 125, so the root is 47.

Can anyone learn these methods?

The methods are short and can be learned in weeks. Competition speed takes years of drill. This post claims no benefit beyond calculation skill itself.

Where Safeguard fits

Safeguard does not do mathematics, and nothing in this post was computed by a Safeguard model; the checks were plain Python. The honest parallel is the checking habit. Champions lose points for every wrong answer, so they verify with cheap checksums like casting out nines and elevens, which catch most slips and miss a few known cases. A software supply chain needs the same discipline: every package you install should be identifiable, checkable against a known good version and traceable to where it came from, because a build, like a fast calculation, keeps no record of how it got its answer unless you make one. Software bills of materials, signatures and reproducible build records are the supply chain's version of a checked result; our post on using SBOMs in incident response shows how that record pays off when an advisory lands.

More in this series

Other posts in our fast calculation methods series:

• Babylonian Mathematics Explained: Base 60, YBC 7289 and Plimpton 322
• Chisanbop Explained: Korean Finger Math to 99, With Worked Examples
• Egyptian and Russian Peasant Multiplication: Doubling and Binary
• Fastest Mental Math Methods Compared: 14 Systems, Same 7 Problems
• Lattice Multiplication, Napier's Bones and the Slide Rule Explained
• Line Multiplication: The Stick Method, Japanese, Chinese or Neither?
• Maya Mathematics Explained: Base 20, Bar and Dot Numerals and Zero
• Trachtenberg Speed System Explained: Rules for 11, 12 and Any Number

Fix it without waiting for upstream

Whatever the vulnerability, most of the risk sits in two gaps. The first is the time between a fix being published and that fix actually running in your production builds. The second, and worse, is the time before any fix exists at all, while a maintainer works out a patch and a release schedule. Waiting through both gaps is a habit, not a requirement. Safeguard gives teams three ways to close them.

1. Autonomous remediation: pull requests, not tickets

When a fixed version exists, Safeguard's Griffin models open a tested, policy gated pull request that moves the affected dependency to the safe version. You can run it on one repository or in bulk across every project that carries the vulnerable version. Remediation covers eight categories, including dependencies, licenses, security posture and findings. Reachability analysis ranks the vulnerable code paths your application actually calls first, so the first pull requests you merge are the ones that matter most. Autonomous remediation ships switched off, with every category set to Manual, so your team decides how much runs on its own and what still needs a human approval.

2. Gold Open Source: the safe version in seconds, free

gold.safeguard.sh is free and needs no login. Search any CVE or package to see affected and fixed versions, CVSS, EPSS, CISA KEV status and a recommended safe version, or use the "Am I affected?" checker. The same data is available through a free read API, README status badges and RSS threat feeds. The safeguard-sh/gold-scan-action GitHub Action can fail a build on a known exploited vulnerability, on any known vulnerability, or never, so a version you have just removed cannot quietly come back in next week's merge.

3. Gold Registry: start from components that are already fixed

registry.safeguard.sh provides hardened, drop in replacements for upstream packages and container images. Every Gold artifact is held continuously to zero exploitable critical or high CVEs, zero malware and a completed license check, with SLSA L3 or higher provenance, a Sigstore signature and a public evidence page you can hand to an auditor. An artifact loses its Gold badge if it regresses. Teams can point their own package directory at the registry so hardened equivalents resolve without changing how developers install anything.

When upstream has no fix yet, or will not ship one, Safeguard first works to get the fix accepted upstream. If the issue is severe or upstream will not move, Safeguard builds a fork, fixes the vulnerability there and publishes a hardened Gold build, so you are not stuck on a maintainer's release schedule. You can also request a hardened, compatibility tested build of a specific package. This fork and patch flow is in early access today, so ask which packages it covers before you plan around it.

One honest limit: when the vulnerable code lives in a vendor appliance, a hosted service or an operating system runtime you do not build yourself, the fix still has to come from that vendor or distribution. The steps above apply to the open source dependencies and container images you build and ship, which is where most teams carry the largest and least tracked share of their exposure.

See how Safeguard would help

These demos show dependency scanning and remediation workflows, not mathematics. You can download Safeguard and follow the CLI docs to try them on your own projects.

This demo is a real CLI session. After installing, safeguard opens the interactive terminal; asked to "fix the lodash CVE in this repo", it reads the finding and the advisory, asks before any command that changes something, edits package.json to lodash 4.17.21, runs npm install and npm test with approval, prints a summary and pushes nothing. The same flow applies to the JavaScript or Python behind a calendar quiz, a timed drill page or a practice app for these techniques.

Click through it at your own pace, or open it full screen.

This demo sets up the Safeguard runner on your own machine, approves it with its device code and scans from the terminal. Pointed at a project folder, such as the small script that checked this post's numbers once it grows real dependencies, the scan lists the packages and versions it depends on and their known vulnerabilities, so you know what to update before you share it.

Click through it at your own pace, or open it full screen.

Quick check

Sources

• Mental Calculation World Cup, organiser's site (recordholders.org)
• Mental Calculation World Cup 2004, results (Annaberg-Buchholz, 30 October 2004)
• Mental Calculation World Cup 2018, results (Wolfsburg)
• Mental Calculation World Cup 2024, general information (Paderborn)
• Mental Calculation World Cup 2024, results
• Mental Calculation World Cup 2024, information sheet (RTF)
• Daniel Timms, Mental Calculation World Cup 2024: Results, worldmentalcalculation.com (September 2024)
• Daniel Timms, Memoriad 2024: Results, worldmentalcalculation.com (14 December 2024)
• Memoriad, official site
• Pesenti M. et al., Mental calculation in a prodigy is sustained by right prefrontal and medial temporal areas, Nature Neuroscience 4(1):103 to 107, 2001
• Lewis Carroll, To Find the Day of the Week for any Given Date, Nature 35:517, 31 March 1887
• Chamberlain Fong and Michael K. Walters, Methods for Accelerating Conway's Doomsday Algorithm (part 1), arXiv 1006.3913 (2010)
• Chamberlain Fong and Michael K. Walters, Methods for Accelerating Conway's Doomsday Algorithm (part 2), arXiv 1010.0765 (2010)
• John Burkardt, DOOMSDAY: day of the week via Conway's algorithm, with the Conway reference
• John Horton Conway, "Tomorrow is the Day After Doomsday", Eureka 36 (October 1973), pages 28 to 31, cited through Burkardt
• Arthur Benjamin and Michael Shermer, Secrets of Mental Math (Three Rivers Press, 2006)
• Bharati Krishna Tirtha, Vedic Mathematics (Motilal Banarsidass, 1965)

Last reviewed: 25 October 2026. Competition formats and record lists change between editions; check the organiser's pages for current rules.