The question is just, is it overall more efficient at broadly "bettering people's lives" in a way they can measure? That's it. If you can show that letter writing is a worthy trade-off by e.g. looking at long-term impacts, I'm sure they'd be for it.
Perhaps you object to the fact that they require measurement. Fair enough, the best things in life are free. But the "bags of grain" (really, "bags of malaria pills") approach is already at a disadvantage to the "give someone a flock of chickens" approach, because the latter tells a story and tugs at our hearts. That doesn't mean it's worse, but I think GiveWell's thesis is, that also doesn't mean it's better.
Grain has some potential negative effects, squashing local food industry, etc, but it's not like domestic bug net companies and a key part of the economy in the same way farming is.
I dunno, I feel like stopping kids from dying is better to focus on first compared to letter writing. But I'm a bit of a Utilitarian, and an aspie, so perhaps I undervalue the human element.
As a side note airlifting in grain in often far more expensive than buying the grain closer to the target area, and causes significant economic harm by putting people employed in farming and the grain supply chain out of work (sometimes the very people the intervention was supposed to help). The main reason we do it is the domestic grain lobby.
> Two-thirds of food for the billion-dollar US food aid programme last year was bought from just three US-based multinationals.
> The main beneficiaries of the programme, billed as aid to the world's poorest countries, were the highly profitable and politically powerful companies that dominate the global grain trade: ADM, Cargill and Bunge.
https://www.theguardian.com/global-development/2012/jul/18/u...
For-profit capture of foreign aid is a much larger problem. For example, remember the news stories after Trump announced cuts to foreign aid saying many retired generals had signed a letter saying foreign aid was vital to protecting our troops? Most forgot to mention that that letter was organized by a trade organization of big defense contractors that regularly win foreign aid contracts (and employ ex-generals)
https://sites.duke.edu/lawfire/2018/02/21/should-we-be-askin...
Furthermore, the more complicated what the charity is doing the more it generally costs to hire competent employees. If a charity wants to hire employees to rigorously evaluate whether their interventions are working, this will also add to overhead.
More controversially, I think the fixation on charity CEO pay is also misguided. They still generally earn less than they could working for similar non-charitable organizations, and being good at CEO-ing is a difficult skill. Hiring a bad CEO can tank an organization. (It's definitely also possible to spend a lot of money on a bad CEO, it's not a guarantee of quality.)
In fact, higher overhead (a negative factor in Charity Navigator's score) correlates with a better Givewell recommendation.
> The mean of the fundraising expense ratio for charities (the money spent on fundraising divided by the total expenses of the entire organization) that earned a gold, silver, or notable rating by Givewell (41 charities) is 0.073, while that of the charities reviewed by Givewell but not ranked well (253 charities) is 0.054 (p-value of 0.05, i.e., statistically significant at the 95% level). The mean of the administrative expense ratio for charities that earned a gold, silver, or notable rating by Givewell is 0.102, while that of the charities reviewed by Givewell but not ranked well is 0.092 (p-value of 0.35). Adding administrative and fundraising together, the mean ratio for charities that earned a gold, silver, or notable rating by Givewell is 0.174, while that of the charities reviewed by Givewell but not ranked well is 0.147 (p-value of 0.11).
(Note that I weakly suspect this person might be misusing p-values, and should be less confident in the causal nature of this correlation. However, that is still sufficient to prove that overhead isn't inherently evil)
http://freakonomics.com/2011/06/09/why-ranking-charities-by-...
Not really the same thing as GiveWell's analysis. To use the first example I checked, compare
https://www.charitynavigator.org/index.cfm?bay=search.summar...
to
https://www.givewell.org/International/charities/Heifer-Proj...