Computers & How Stuff Works

A bit has two possible values, 0 or 1. An unsigned binary number uses powers of two as place values. In four bits these are 8, 4, 2 and 1. A byte contains eight bits. With n bits there are 2ⁿ distinct patterns; an unsigned n-bit integer ranges from 0 to 2ⁿ − 1.

The pattern 1010 represents ten when interpreted as an unsigned binary number. The same stored pattern can have a different meaning under a different encoding. Text requires a character encoding; images use representations of pixels and colour; sound can be sampled and quantised. A computer needs conventions, not merely flashing lights.

Adding binary numbers uses carrying just as decimal does, but each column carries at two rather than ten. A result outside the available number of bits causes overflow in a fixed-width unsigned representation. The virtual four-bit sender shows only unsigned place value; it does not implement encryption or all computer data types.

A processor executes machine instructions. In a conceptual fetch–decode–execute cycle, it fetches an instruction from memory, decodes its operation and carries it out. Registers hold small amounts of data used immediately. The arithmetic and logic unit performs arithmetic and logical operations; the control unit coordinates instruction execution.

RAM stores data and instructions currently in use and is usually volatile: contents are lost when power is removed. Secondary storage keeps files for later use. Storage capacity is different from processing speed. A program loaded from storage can use RAM while the processor executes instructions; it does not make the CPU “remember” as a person does.

Real processors overlap operations, use caches and have complex designs, so the simple cycle is a teaching model. Explain input, processing, output and storage using a familiar task such as adding two numbers. An operating system manages resources and provides services for applications. Programs and hardware are different parts of the system.

A network connects devices so they can exchange data. Data may be split into packets containing payload and control information, such as destination and sequence information. Packets can take different paths and must be arranged appropriately at the receiver. Bandwidth is a rate capacity, not the same as how long one message takes to arrive.

Noise or faults can change transmitted bits. Error-detection methods add information so some changes can be noticed. In even parity, the added parity bit makes the total number of 1 bits even. It detects any odd number of bit flips, but can miss an even number. Acknowledgements and retransmission can help recover from detected errors.

Encryption addresses confidentiality by transforming data under a scheme and key. It is not the same purpose as a parity check. Safe digital behaviour also includes trusted updates, careful access controls and avoiding disclosure of personal data. This hall’s light-signal activity is a representation exercise, not a secure communications system.

A bitmap image represents a grid of pixels. Each pixel’s value encodes colour under an agreed format. Resolution refers to the pixel grid or sampling detail, while colour depth is bits per pixel. For an uncompressed simplified image, raw data size in bits = width × height × bits per pixel. Real files also contain metadata and may use compression.

Digital sound records samples of a changing signal. Sample rate is samples per second and bit depth controls the number of quantisation levels. For a simplified uncompressed recording, size in bits = sample rate × bit depth × duration × channels. More samples or more bits require more storage; they do not automatically improve poor source material.

Lossless compression allows exact recovery of the encoded data; lossy compression discards some information to reduce size. Choose a representation according to purpose and constraints. A medical system, a text document and a shared music clip can have different requirements. Keep units clear and distinguish bits from bytes.