Semantically, you are right.

But, that is not the issue Ovid raised in Recap: The Future of Perl 5.

In his proposal, Ovid seeks to reduce the number of required steps and simplify the syntax in Perl 5's OO system.

As to why I said Python's OO system seems to have changed since Larry stole from it, see the Python 3 example:

class Dog:

    # Class Attribute
    species = 'mammal'

    # Initializer / Instance Attributes
    def __init__(self, name, age):
        self.name = name
        self.age = age


# Instantiate the Dog object
philo = Dog("Philo", 5)
[download]

__init__ is called with an already "blessed" ref to an already allocated container. __init__ only sets values to the objects's fields.

The 1-to-1 translation to Perl would be (using signatures to reduce "noise"):

package Dog;

our $species = 'mammal';

sub __init__ ($class, $name, $age)
{
    my $self = {};
    $self->{name} = $name;
    $self->{age} = $age;
    bless $self, $class;
}

package main;

my $philo = Dog->__init__('Philo', 5);
[download]

Even if we make __init__ more Perl-ish (still using signatures):

sub __init__ ($class, $name, $age)
{
    my $self = bless {name => $name, age => $age}, $class;
}
[download]

Ovid would still point out that Perl is requiring the programmer to perform 2 extra steps: Allocate the container and create a blessed reference to the container.

I don't know which version of Python Larry stole the OO system from, so I don't know what Larry changed. But, at least to me, Perl 5's classes look more like Smalltalk classes than Python 3 classes.