celery chaining issue

tasks.py

@celery.task
def first():
    # generates an iterable
    return 1, 2, 3


@celery.task
def second(thing):
    """
    Takes a single item from the iterable from `first`, does some computation,
    and returns an iterable.

    There will be one `second` task for each item in the iterable returned by
    `first`
    """
    return [data for data in range(thing)]


@celery.task
def third(data):
    """
    Takes a sindgle item from the iterable from `second` does some computation,
    and returns an iterable
    """
    return 1  # result


@celery.task
def fourth(results):
    """
    Takes the results of all `third` tasks, and does some processing on them.
    """
    print "####################################"
    print 'results', results
    print "####################################"
    return sum(itertools.chain(*results.iterate()))


@celery.task
def map_to_second(things):
    """
    This concept works pretty good if there is only one level of 'chording',
    but in my situation, I have two.  How do I have the chord from the top level
    """
    return celery.chord(
        celery.chain(second.s(thing), map_to_third.s()) for thing in things
    )(fourth.s())


@celery.task
def map_to_third(list_of_data):
    """
    This function doesn't have enough 'context' to feed the results into
    `fourth`.
    """
    return celery.group(
        third.s(data) for data in list_of_data
    ).apply_async()


@celery.task
def doit():
    return celery.chain(
        first.s(),
        map_to_second.s(),
    ).apply_async()