"Can you make this essay shorter?"
Yuki showed her report. It exceeded the word limit.
"You need compression," Aoi said.
"Compression?"
"Data compression. Technology to reduce size without losing information."
Mira approached and wrote in her notebook. "Lossless and lossy compression."
"There are two types."
Aoi explained. "Lossless compression can be perfectly restored. Like ZIP format."
"Lossy compression loses some information but can reduce size significantly. Like JPEG."
Yuki thought. "Which is better for a report?"
"Lossless to preserve content. But in your case, lossy is also an option."
"Lossy?"
"Delete unimportant decorative words. Make redundant explanations concise."
Mira showed an example. She pointed at Yuki's text and drew red lines.
"This adverb isn't necessary for the meaning," Aoi explained.
"True..."
"The basis of compression is removing redundancy."
Yuki asked. "Redundancy means waste?"
"Not necessarily. There's redundancy needed for error correction."
"But if there's a word limit, you have to cut."
Aoi continued. "Compression uses statistical properties of data."
"Statistical properties?"
"Encode frequent words short, rare words long. Like Huffman coding."
Mira drew a tree structure in her notebook. A binary tree spreading out.
"Frequently used characters are closer to the root. So their codes are shorter."
Yuki understood. "That's efficient."
"But," Aoi warned. "Compress too much and meaning won't get through."
"The limit of lossy compression?"
"Right. JPEG also shows block noise if compression rate is too high."
Yuki reviewed her report. "It's hard to know how much to cut."
"That's the tradeoff with information content."
Mira took out a new notebook. It said "Compressing emotions."
"Compressing emotions?" Yuki asked curiously.
"Humans also compress emotions to convey them," Aoi explained.
"Packing long feelings into short words. How much information is compressed into the two characters 'I like you.'"
"True..."
"But the receiver can't always decompress it correctly."
Yuki thought. "Decompression means decoding?"
"Right. Restoring compressed data. But because it's lossy compression, it doesn't fully restore."
"So misunderstandings arise."
"Words are always lossy compression. You can't convey all of your thoughts."
Mira nodded. Then wrote. "So completion is necessary."
"Completion?"
"The receiver fills in missing information from context. This is also part of decompression."
Aoi supplemented. "Like super-resolution in machine learning. Estimating high resolution from low-resolution images."
"Filling in with imagination."
"Right. That's why different people receive the same words differently."
Yuki looked at her report. "Then I'll compress too. Trusting readers will decompress."
"Good pragmatism."
Mira wrote again. "Perfect compression doesn't exist."
"Shannon's source coding theorem," Aoi said. "Can't compress below entropy."
"There's a limit."
"But we can strive to approach that limit."
Yuki started writing. Cut, cut, leaving only the essence.
"How's this?"
Aoi read it. "Good compression. Meaning is preserved."
Mira smiled and gave a thumbs up.
"Encoded feelings will someday be decompressed by someone."
Yuki murmured.
Aoi answered. "That's the hope of communication."
Compression and decompression. People live repeating this every day.