Learning gamedev is hard. It combines multiple creative and technical disciplines, many of which have punishingly steep learning curves. So motivation is often a problem early on. It's easy to feel defeated every step of the way, and giving up is the norm.
Having a "Dream Game" can be useful for learning in this regard. A Dream Game can provide that boost of motivation you need to get through the hopeless times. You can channel your passion and vision into the determination to learn. Conversely, working on small games can be massively demotivating. I don't want to make Breakout or Flappy Bird, I want to make Chrono Trigger. I want to flex my creativity and my game design intuitions, I want to build something amazing. And here I am, struggling so hard to make a game I don't even care about. What's the point.
This is why it's so common for new devs to work on huge dream projects as a vehicle for learning the craft. And as long as you're learning, that's great. But you must understand that this game is your sacrificial lamb. You will never finish it. Granted, some devs do manage to succeed at the "learn while I build my Dream Game" strategy - Stardew Valley comes to mind. But almost everyone else will fail, and for a few reasons.
The first is that you cannot size a project unless you know what it takes to finish. How long do you want to be working on this Dream Game? 4 years? Okay, so how are you supposed to know what a 4-year game looks like? Unless you've finished a game before, you cannot know. You've likely misjudged the effort it'll take by an order of magnitude.
Secondly, finishing a game requires its own set of disciplines. There's so much to learn about putting out a commercial product. Have you considered graphics options? Resolution, framerates, vsync? Audio settings? Accessibility? Localisation? Do you know what it takes to QA a game? Do you know how to configure builds and depots on Steamworks? Do you really want to be encountering all this stuff for the first time at the very end of a giant 6-year project?
Thirdly, it's easy to get bogged down and stymie your learning when you have such a large game to make. You can lose six months working on animated cutscenes before you've even learned how to make a pause menu. Everything gets dragged out, and inevitably you end up remaking huge chunks of the game as you come to realise that your early work was amateur trash - as is to be expected.
Lastly, dream projects provide far too many distractions from actually learning game developement. Writing lore. Drawing concept art. Dialog. GDDs. These are low value tasks, but they're so much fun that you can easily spend all your time on them. And before you know it, it's been weeks since you opened your project file.
Finishing and releasing small, simple games is the remedy to all these problems. Make a warioware-style minigame, make tetris, make a shmup, make a 10-minute avant-garde experimental FPS hallucination simulator. It doesn't matter, just make a game that people can download and play, then make another one.
You can still build your Dream Game. Just later. If anything, it deserves to be tackled properly, by a dev who's got some real holistic experience under their belt.
make small games. if you don't have the skills to hack out a polished Pong clone in a weekend, your 6-year dream game project is in serious trouble