Forgiving the problem of time dilation and the consequences it has on the expectations of conscious things...

You can't just accelerate 1g for 3 years. What are you burning? A chemical rocket can easily achieve 1g. For about 5 minutes. A nuclear rocket is great, you can get 1g easy enough, but you're shoving hot hydrogen out the back end to get that thrust.

The Rocket Equation teaches us that to get arbitrarily close to c, you need an exponential amount of your ship's mass to be propellant.

I recommend this website to everyone in this thread that hasn't read it yet

http://www.projectrho.com/public_html/rocket/

Edit: moreover, the problem of an EM deflector, or any kind of deflector, is that no matter what, no matter what, you must deal with the kinetic energy of the thing that just hit your deflector. It will (a) slow your ship down, and (b) cost you more energy, e.g. from your generator, to deflect it than it imparted on your vehicle. The energy numbers here are staggering. We have trouble theorizing how a relatively uncontrolled terawatt rocket might work, but imagine a terawatt reactor on your ship somehow powering a terawatt EMfield in front on your ship.

And then you have to worry about dust that's not charged and thus goes right through EM fields.

Indeed, and the benefit of traveling at 0.995 c is that a container of information doesn't suffer from degradation by the inverse square law as it travels over a large distance. It might actually cost less energy to transport information to far flung destinations that way, interstellar snail mail if you will. Some amount of information has to accompany the robot anyway, at the very least, its initial operating instructions.

Of course whoever can get there at 0.996 c can take advantage of a market arbitrage. ;-)

> a container of information doesn't suffer from degradation by the inverse square law

Neither does a well-focused laser?

EDIT: apparently it does, unless you manage to focus your laser very far away. Not sure how practical that is.

1kg matter at 0.1c has a kinetic energy of 4.5x10^14 J

It is certainly very expensive to spend this amount of energy. But is it also extremely challenging to even extract this out of an engine that has to accelerate itself.

1kg matter at 0.995c has a kinetic energy of 8x10^17 J

This is the amount of enery you get if you convert 9kg in pure enery. That seems impossible to reach, even with an anti matter drive.

A 9:1 ratio of fuel to mass doesn't sound particularly bad. Are there known theoretical limits to extracting the full E=mc^2 worth of energy from an object?

No, you can do this with an antimatter rocket.

Good luck capturing and containing antimatter and feeding/containing the explosion though. It's going to be practically impossible to get a 90% fuel ratio from an antimatter reactor, since literally 45% of your ship will be antimatter, by mass.

You can also do this by feeding a black hole with ordinary matter and using Hawking radiation for propulsion. That is probably more practical than carrying around large quantities of antimatter.

https://en.wikipedia.org/wiki/Black_hole_starship

If you watch PBS Spacetime on the 'Kugelblitz' idea, it seems that one way to go about it (over the course of a couple hundred years) would be to convert Mercury into a solar energy collecting swarm by use of Von Neumann probes and essentially harvest much of what the sun has to give. That kind of energy apparently would suffice to create tiny black holes (simply by redirecting enough energy to a focused point), which is I think the theoretically most efficient battery (or star system destroying bomb, depending on how you go about it...).

I... can't.... Half of me recoils in horror at the impracticality, but the other half of me has a good enough armchair sense for the practicality of a high fuel ratio antimatter rocket, so I can't say you're wrong :P

You need to accelerate the fuel as well. And for a reaction engine, you need to carry reaction mass that you accelerate. And at some point you want to decelerate again.

Would be interesting to see whether this is at least theoretically feasible.