But think of it this way: Bascially I'm losing the opportunity cost of 10,500 dollars, spent very early in the accumulation process. I've already accumulated nearly 5000 in gains on that money. That is 5000 dollars that needs to be excluded from my FIRE number, because I can't access the money. Think of how much will be there in another eight years...
Your youthful recency bias is showing, not just in your tax-law perspective, but perhaps in your investment expectations as well. If you were 15 years older, investing in 1998-1999, you also would have seen your $10.5k investment grow by $5k. Then, think of how much would have been there in another
eight years! Hey, thankfully our time machine allows us to look into the "future" in that case for the answer: about $15k total, same as 8 years before. And soon to drop to ~$9k, an amount lower that your initial $10.5k investment. There you go, problem solved! You don't have to worry about your "trapped" earnings, because you have no earnings at all.
Yes, that's an extreme example, but the point is that there is no guarantee that your Roth contributions will end up as a meaningless fraction of your overall Roth balance.
I am inclined to think that 6% of dividends is fairly insignificant as far as expenses go though: with dividend yields on VSTAX at 1.7%, this amounts to an effective .1% expense... that can be written off of federal taxes!
Well, to make a decision here, you at least need to do the math, so I'll help get you started. A 0.09% "expense" over a 30-year period reduces your portfolio value by 2.67%. That's a lot more than 0.09%, though still not a giant number. But say the federal dividend tax rate, which has changed several times in the last decade alone, goes from 0% to a still-historically-low 15%. Then you're looking at something more like a 0.37% "expense", which adds up to a 10.52% portfolio-reduction over 30 years. That's more paid in dividend taxes than if you just paid the 10% tax for early withdrawal from the Roth. If VTSAX's dividend rate increases, or you make more money than you expect, or if dividends go back to being federally taxed at ordinary income rates and the tax brackets change, or if you move to another state with higher taxes, or your state raises its taxes, than the portfolio reduction will be even greater than that 10.52%.
So if you can plug in numbers for the likelihood of any of those things happening, you can come up with an "expected value" for your portfolio-reduction due to dividend tax exposure over its lifetime. That's hard to do, but I'd say you can be fairly confident that your reduction will never be less than 2.67% over 30 years, and there is, I dunno, maybe a 30% chance of any or all of those things turning it into a ~10% reduction?
Is that reduction a fair price to pay for the ability to access money earlier than 59.5, an age where you're likely to need money for 40 more years anyway? Maybe it is, but that's at least the framework you should be using to make the decision.
Finally, remember, just like in school, the principal is your PAL! Not your PLE. Who the hell wants a PLE?