Add your Twitter credentials to your environment, run this and notice how the second request yields only one tweet? Try running again with a sleep between them (uncomment line 27) and notice how the second request finds results this time? This seems to be a bug, or a new "feature" intended to prevent using the API to walk backwards in search res…

max_id_bug.py

#!/usr/bin/env python

# you'll need to install python's oauth2 for this to work

import os
import json
import time
import oauth2

ck = os.environ.get('CONSUMER_KEY')
cks = os.environ.get('CONSUMER_SECRET')
at = os.environ.get('ACCESS_TOKEN')
ats = os.environ.get("ACCESS_TOKEN_SECRET")

consumer = oauth2.Consumer(key=ck, secret=cks)
token = oauth2.Token(at, ats)
client = oauth2.Client(consumer, token, timeout=60)

q = 'dconstruct'

print "request #1"
resp, content = client.request('https://api.twitter.com/1.1/search/tweets.json?count=100&q=%s' % q)
for s in json.loads(content)['statuses']:
    print s['id_str']
    last_id = int(s['id_str'])

# time.sleep(20)

print "request #2"
resp, content = client.request('https://api.twitter.com/1.1/search/tweets.json?count=100&q=%s&max_id=%i' % (q, last_id))
for s in json.loads(content)['statuses']:
    print s['id_str']