It was based on an intuitive idea: if teenagers see how bad jail is, they'll want to do anything they can to avoid it. That turned out pretty poorly - they instead built social connections with criminals, and were substantially more likely to get involved with crime themselves.
[1] http://blogs.ei.columbia.edu/2010/07/01/the-playpump-what-we... [2] https://www.upi.com/Defense-News/2003/04/02/UNICEF-Rations-b...
For one thing, the fact that it is tax deductible itself results in it being the focus of self-serving tax dodges that aren't really charitable. If charities remained tax exempt but donations weren't tax deductible, the abuse incentive would be removed.
There's also, of course, well meaning but counterproductive charities, but the subsidy also probably increases the share of donations that goes to those, since it reduces the incentive for donor diligence.
https://www.livemint.com/Opinion/b6rCZ0tubpb0GIYfq7hnBM/Weak...
It probably doesn't take much work to ignore the latter, as long as you ignore emotional appeals and use a little bit of reason instead. But it takes some work if you want to go up the spectrum, if you want to find the more effective charities. If saving lives is your primary concern, then GiveWell does great jobs in rating charities by their effectiveness (in dollars per life saved). If you're interested in different types of social impact, then you'll likely have to do this work yourself.
Even with good intentions, predicting outcomes is difficult. When you donate to a charity(or any organization), you are giving them more influence while delegating decision making. The best option seems to be: find something you really care about and get your hands dirty.
Some also feel that donating has a risk of creating dependency.