Rather galling, considering that our result was timed first (by one second).
You can send ghash.io your thanks on that one. Their connectivity with the majority of nodes/pools is downright malicious in how bad it is.
How can ghash.io negatively influence another pool's orphan rate? I can see how being well connected can reduce their own orphan rate, but how can they increase someone else's orphan rate?
Most pools are in at least decent datacenters, which happens to mean they have low latencies with the other pools since they're both close to backbone bandwidth providers. This reduces overall network orphan rate because the average time between block notifications between the majority of the network hash rate is significantly lower than what it would be between people just running on random local ISPs. At the same time, GHash.io has abysmal latency and/or absolutely no peers setup between pools, or they're purposely ignoring competing blocks sometimes based on the sheer number of orphan races they compete in. This is based on admittedly more casual observation rather than deep analysis, but I know other pool ops have been seeing more losing orphan races than usual as a result of ghash.io.
Hopefully ghash.io is not testing out the selfish mining strategy...