When the p function is used to print out an object, it may give an ID, and it is different from what object_id() gives. What is the reason for the different numbers?
Update: 0x4684abc is different from 36971870, which is 0x234255E
>> a = Point.new
=> #<Point:0x4684abc>
>> a.object_id
=> 36971870
>> a.__id__
=> 36971870
>> "%X" % a.object_id
=> "234255E"
The default implementation of inspect calls the default implementation of to_s, which just shows the hexadecimal value of the object directly, as seen in the Object#to_s docs (click on the method description to reveal the source).
Meanwhile the comments in the C source underlying the implementation of object_id shows that there are different “namespaces” for Ruby values and object ids, depending on the type of the object (e.g. the lowest bit seems to be zero for all but Fixnums). You can see that in Object#object_id docs (click to reveal the source).
From there we can see that in the “object id space” (returned by object_id) the ids of objects start from the second bit on the right (with the first bit being zero), but in “value space” (used by inspect) they start from the third bit on the right (with the first two bits zero). So, to convert the values from the “object id space” to the “value space”, we can shift the object_id to the left by one bit and get the same result that is shown by inspect:
> '%x' % (36971870 << 1)
=> "4684abc"
> a = Foo.new
=> #<Foo:0x5cfe4>
> '%x' % (a.object_id << 1)
=> "5cfe4"
Note: The details about object_id were correct at the time (2010), but aren't anymore after newer Ruby versions have decoupled object_id from memory addresses. Now the answer is more along the lines "because object_id is generated on demand". See comments.