What's the benefit that justifies a 2-3x increase in code size?

I agree that it doesn't necessarily make life any easier, and the code increase (line count-wise) is alot less extreme if you ignore the Tangram schema setup and the comments and the line breaks for readability. But the benefit is that its a tool for 'object persistence'. Which puts Tangram and EZDBI in entirely different categories anyway. But I do appreciate the examples of both :)

Update:Re: Dominus's reply - I also said 'ignoring the Tangram ... setup', which is everything up to the 'package main'. Which you can count or not. I figure there's an argument for not counting it since its a constant that, once set up, you can do as much inserting/deleting/updating/selecting as you like. So IF you don't count it, then the amount of code is not much greater.

Since the original program didn't have any objects, I don't see how that could be a benefit in this case.

That was my point, the two modules are for entirely different things :) Making DBI EZ or storing objects in a database. And how can you say its not a benefit in this case just because your script didn't have any objects. I kind of think of Harry Potter as an object :)
Besides, if you're going to make that sort of argument, then I submit that this script is far superior:

use HarryPotter;

InsertHarryPotter();
IsPotterInTheHouse();
PrintHarry() for SelectPotter();
DeletePotter();
SelectPotter() and DamnPotter();
[download]