I've won a few ladders against IGowin and wondered why it couldn't see the obvious. I know in chess endgames it's often easy for humans to see outcomes a dozen or more moves ahead because you can "eyeball it" without needing to calculate every move. Is this limited horizon feature characteristic of Go programs too ? They can't read ahead very far at all and are therefor bad at ladders ?

Igowin is very weak. Better programs usually have special routines for ladders (as better chess programs have endgame databases so that they play a perfect endgame).

I played Many Faces some time ago and it also made a mistake with a ladder.

Here is the game, I'm white and Many Faces is set as 2k:
The ladder starts at

Do you know which iPad program plays the strongest most human like game ?

Not the way to frame your question.

The proper way is to refer to Program X version Y (not even any version of a specific program and certainly not all possibole programs on all possible computers).

Note that this also takes care of the "some time ago" matter which is very important as a year or two ago is a fairly long time in the development of go programs. Specifying version more useful than specifying when. It doesn't matter if your experience was yesterday in real time. If that experience was with a version three years old then that's equivalent to an experience three years old and not relevant to what these programs might or might not be able to do today.

And please note: This forum has an area for computer go discussion.