Have you experimented with the approach of trying to use positive motivation on yourself to donate more instead of negative motivation? In other words, think of life like a computer game where the objective is to save as many lives as possible. You increase the rate at which your score grows by levelling up in your career and finding clever ways to subsist on less money, and you can periodically collect bonuses by hosting effective altruism meetups. You don't necessarily have to stress yourself out trying to eke out the very highest score possible... then the game would stop being fun, and besides, stress will make it harder to get the high score anyway (life is a marathon, not a sprint--and it's very complicated and full of potential shortcuts, so you're probably better off using your mental energy to look for shortcuts than risking being penny-wise and pound-foolish).

There are a lot of people who are trying to figure out what their purpose is and what they want to do with themselves. Making money is an interesting problem, and when it comes to doing good for the world, there aren't many careers whose direct effects compare with donating lots of money.

(These are just thoughts; it may be that your method of motivating yourself tends to be a more reliable way to produce donations in practice. But I do wonder if my method would work better for folks who have heard fungibility arguments and don't seem to take them seriously, at least in terms of framing all donations as points earned even if they donate a relatively small amount and only earn a small number of points.)

I could go through my life constantly weighing every purchase against the good it could do, but this would make me miserable.

You could be weighting less frequently (there is little value in reevaluating essentially the same decision too often), and it's not axiomatic that considering the decision would make one miserable. The tradeoff is real, so if at all possible (without sacrificing the correctness of decisions about it) one should learn to become comfortable with its existence. (I expect it isn't possible for some people, which is where workarounds become relevant.)

Specifically, I can donate to the Against Malaria Foundation, which distributes anti-malarial nets. The main effect is averting deaths of children who will probably go on to live around 30 years once you take into account other things they might die from. This comes to about $75 per additional year of life.

And those life-years have no other financial costs and benefits? Or does the labor market participation of these saved children exactly make up for that?

Also, does the death rate from malaria have no effect on the birth rate? E.g. replacement pregnancies?

Personally, I use the old "tithe" heuristic, with half going to political causes and half to more conventional charity. I doubt it's optimized, but my financial situation is uncertain enough going forward that I don't want to go too crazy with donations - I'm accumulating a lot of savings instead, and those can always be donated later. But I want to work for myself first, mostly out of simple selfishness, but partially because that's the society that tends to do best at actually making people off, and I don't want to wreck that social norm.

So how do I accept that other people need my money more without giving up on being happy myself?

Personally, I've accepted the fact that I'm not perfectly altruistic...even though I might wish for strangers to be perfectly altruistic. Even if you are a utilitarian, that doesn't mean your "utility function" is utilitarian.

I might estimate that giving $100 to the AMF does about as much good as being vegan for a year, so if I decided to go ahead with being vegan I would decrease my annual donations by $100 and allocate another $100 to spend on myself

If you had a utility function (which you don't, really, since you are human) it wouldn't work this way. If Veganism generate positive utils, and $100 donations generate positive utils, you would do both, rather than one or the other...unless there was some weird interaction effect whereby donating $100 made veganism less satisfying or vice versa. I suppose technically guilt alleviation does create an interaction effect...but you created this interaction yourself, by mentally deciding that one good deed makes up for not doing another.

But my point is that if you take messy emotions like guilt out of the equation, "trading" one good deed for another doesn't really make sense. If you instead accept that you are selfish and disregard guilt, the thought process might look more like this:

1) I'm selfish and want resources for myself and my loved ones. However, I'm also altruistic, and there are diminishing returns to spending on myself - once I control at least X resources, it pleases me better to give all the rest of my money to these other people.

2) Eating meat pleases me, but killing animals displeases me more, therefore I will not eat meat.

The utility of each separate action is considered separately - without the confound of guilt atonement, there is no interaction between these two decisions.

Of course, there is no rule saying that you have to behave as if you didn't have emotions such as guilt...I just thought I'd point out how one might balance selfishness against altruism if guilt weren't a factor, since that knowledge might help to encourage altruism without guilt as a motivation.

EDIT: just realized we had this conversation before...oops...