Picture sorting a hand of cards. You pick up one card at a time and slide it into place among the cards you already hold, which stay in order: that is insertion sort. Bubble sort works differently. Walk along the row, compare each neighboring pair and swap them if they are out of order. Sorting [5, 1, 4, 2], the first pass swaps 5 past every other card, giving [1, 4, 2, 5]: the largest value has 'bubbled' to the end. The next pass fixes the second largest, and so on.
Methods like these compare nearly every pair, about n2/2 comparisons, so doubling the list quadruples the work. Merge sort is smarter. Split the list in half, sort each half, then merge the two sorted halves by repeatedly taking the smaller front item. That costs about nlog2n comparisons: for 1,000 items, about 10,000 instead of about 500,000.
Every sort needs a key and a direction. Sorting students by name gives a different list from sorting them by test score, and a leaderboard needs descending order. When two items share a key, a stable sort keeps them in their original order. That matters when you sort by one field after another: sort by name, then stably by grade, and the students within each grade stay alphabetical.
Sorting is not needed for every question about order. Finding the single highest score takes one pass of n−1 comparisons, with no rearranging at all.
Think of it like this. Lining up a class by height for a photo is sorting. Neighbors can keep swapping until no pair is out of order (bubble sort), or the class can split in two, line up each half, and merge the two lines by sending forward the shorter front person each time (merge sort). A computer cannot see the whole line at once; it compares two people at a time.
Core idea. Different sorting algorithms have different efficiency trade-offs and work better in different situations.