The world if floating point math just worked

Description

A classic 'The World If' meme format, which depicts a gleaming, futuristic, utopian city with sleek, towering skyscrapers and advanced infrastructure. Superimposed over this idyllic scene is white, bold text. The top text reads 'THE WORLD IF', and the bottom text presents a simple mathematical equation: '0.1 + 0.2 = 0.3'. The joke is a deep-cut reference for programmers to the inherent inaccuracies of floating-point arithmetic in most programming languages due to the way decimal numbers are represented in binary (as defined by the IEEE 754 standard). In reality, 0.1 + 0.2 evaluates to a number like 0.30000000000000004, a common source of bugs, especially in financial and scientific applications. The meme humorously suggests that this fundamental, persistent quirk of computing is the primary obstacle preventing humanity from achieving a perfect, utopian society

Comments

Anonymous ★ Top Pick The correct way to check if 0.1 + 0.2 equals 0.3 is to see if the absolute difference is less than epsilon. Or, as we do in the real world, multiply by 100 and work with integers

Anonymous ★ Top Pick

The correct way to check if 0.1 + 0.2 equals 0.3 is to see if the absolute difference is less than epsilon. Or, as we do in the real world, multiply by 100 and work with integers
Anonymous

In the universe where 0.1 + 0.2 == 0.3, we shipped the fintech stack in one sprint - no BigDecimal, no ε, only the PM’s timelines still overflow
Anonymous

After 20 years in the industry, I've accepted that we'll achieve cold fusion before JavaScript properly handles decimal arithmetic - at least BigDecimal is finally in Stage 2, only took us three decades to admit that financial calculations shouldn't rely on binary approximations of base-10 numbers
Anonymous

Ah yes, the IEEE 754 fantasy - where we've finally achieved a civilization advanced enough to represent 0.1 in binary without an infinite series. In this utopia, financial systems don't need BigDecimal, JavaScript developers sleep soundly at night, and 'use an epsilon for comparison' is just an ancient proverb from the before-times. Truly, we'd have colonized Mars by now if only those three bits of mantissa had cooperated
Anonymous

Dev heaven: exact decimals by default, sparing us epsilon dances in every prod-facing aggregation query
Anonymous

In a universe where 0.1 + 0.2 === 0.3, I’d finally delete epsilonCompare(), half the flaky tests, and three ADRs debating floats-for-money - Finance would still argue about rounding
Anonymous

If 0.1 + 0.2 ever equaled 0.3, we’d delete closeEnough(), retire BigDecimal, and stop explaining to Finance why their budget is 0.30000000000000004
@Sokolovskiy01 5y

But...
@average_meni_na_drugu_enjoyer 5y

in reality it 0.3000000006?
@RKRohk 5y

== tho
1. @JoseAngelSanchez 5y
  
  True
@mashkachan 5y

i dont understand this joke
1. @deerspangle 5y
  
  Using floating point numbers, addition like this generally leads to a very small error, so it would equal 0.3000000006 instead of 0.3
  
  @mashkachan 5y
  
  Ok, thanks
2. @Supuhstar 5y
  
  If you try this in a programming language which uses normal IEEE floating-point numbers (basically all programming languages except Mathemetica and the like), then: 0.1 + 0.2 = 0.30000000000000004
  
  @mashkachan 5y
  
  Its can be problem with a super accurate calculations
  
  @Supuhstar 5y
  
  and not-so-accurate, as this simple one shows
  
  @devotus08 5y
  
  but why it is in that way?
  
  @Supuhstar 5y
  
  Because you're thinking in decimal, and the computer is thinking in binary. If you wanted to use only binary fractions, it'd work out better, but not always accurate. This is all due to the format that IEEE floating point numbers are stored in. They're all similar, but let's focus on 32-bit floats. They have one bit for whether it's negative ("sign"), 8 for how big it is (exponent), and 23 for how precise it is (significant, or mantissa, or fraction). In order to calculate what it means and how to make one, you have to use the formula in this picture. If you go through by hand, you'll see that it's impossible to represent 3/10
@GTRst 5y

JavaScript devs left the chat
@Supuhstar 5y

OOF
@nuntikov 5y

Use decimal
1. @Supuhstar 5y
  
  Exactly. Never use floating-point primitives for anything but estimation
@energizer91 5y

https://0.30000000000000004.com

Use J and K for navigation

The world if floating point math just worked

Description

Categories

Comments

Related Memes

Deterministic Randomness Exposed

Any Sophomore Can Monad This

SMBC: Consciousness Is Just a Bandwidth Issue With 4,000 Agents