There are still miners out there with old versions of the software and running the defaults. Thus, if there's not too much of a backlog, it's quite possible zero fee transactions will get included.
So if zero fee transactions are actually a problem, perhaps it is better that we do have a bit of miner centralization so we have professional miners who actually pay attention to what they are doing instead of a bunch of amateurs not acting in their rational self-interest as was supposed to happen.
Don't amateurs mainly hash for pools? It seems hard to believe there's any significant hashrate operated by hobbyists who actually run their own bitcoind for mining.
Good point. And mainly so I think. In fact, some of the offenders seem to be large pools. Antpool and slush appear to be still mining with a 750000 soft limit so it would be interesting to see if they were mining 0 fee transactions also.
I send 0 fee transactions all the time and so far they've always* gone through, so there's definitely still a decent percentage of miners that accept them. The only time I pay a fee is if I'm using a web wallet or phone wallet where the fee is built in, or if I'm sending coin to an exchange to make a trade and time is of the essence.
* (I did have 1 unconfirmed from a test during the mtgox malleability fiasco, but it actually went through, just my client thought it didn't)
What's the longest wait you had for a transaction to go through? When the spam attacks were running people who paid fees were waiting over half a day, and the backlog took a week to clear. Did you try sending any zero fee transactions while those attacks were ongoing?