Sometimes the fastest way to make progress is to take everything off the breadboard and just have a board, a USB cable, and a serial monitor. No LED, no sensor, no wiring. You can still write meaningful sketches — and the constraint forces clarity.
What you can do with nothing wired up
You can profile. Print millis() in your loop and see how fast it really runs (often hundreds of thousands of times per second). Add a delay and see exactly how it affects rate. You can test serial-only protocols — command parsers, line interpreters, EEPROM persistence. You can rehearse algorithms — bitwise ops, fixed-point arithmetic, lookup tables — and verify with prints.
You can also learn about your sketch's footprint. Serial.println(__DATE__); and Serial.println(F("hello")); teach you about Flash vs SRAM. Serial.println(sizeof(int)); tells you it's two bytes on AVR, not four. These small interrogations build the intuition that makes you fast later.
Why this matters
Real embedded projects often start with this same bareness: get the host comms working before anything else. If you can't print, you can't debug. If you can't debug, you can't fix bugs. A bare Uno + Serial is the developer's equivalent of "make sure you can talk to it" — an unromantic step that prevents weeks of pain later.