Another parable comes to my mind.

• You won't come to work, and boss comes and realizes. You won't be paid. (Short round, end of round)
• You leave work for 2 hours, return at luchtime, boss comes in a while, and colleagues remember and tell him. You are paid less. (Some time before the end of round.)
• You are late to work in the morning, boss comes in the afternoon, no one remembers, you're fully paid. (Beginning of a long round.)

Well, just for fun... :-)


Are you? Are you mining solo? :-)
OK, let's modify the above example, when you don't feel yourself "employed" by Slush. ;-) If you're self-employed, you get money for some goods or services.
Then, when you don't deliver your goods or services, you're not paid. OK now? ;-)

 For those with large amount of hash power, setting a higher difficulty will reduce the network bandwidth consumed by submitted shares.  You don't have enough hashpower for it to matter.   I wouldn't worry about it. 

So just curious at what point would you need to set a higher difficulty? How many Mhash/s are we talkin'? I know what the website suggests according to Mhash/s but is this realistic? What have others experienced?

all increasing the difficulty does is cut down on the network traffic. mining uses so little bandwidth that i wouldn't worry about it. besides, i think the pool raises your difficulty automatically if you set it too low... or maybe that was a different pool. it all blurrs sometimes.

I was hoping it was going to be corrected before it got confirmed but I guess by the low amount of others having problems with 18994 it must have been on my end?

Not really.
The difficulty value you set in your client tells your client to send only shares of that difficulty (and higher) to the server. And it tells the server that you will be doing so.
The server will than count each your difficulty 2 (or higher) share as two difficulty 1 shares (when you set the difficulty to 2, of course). Your network traffic is then lower, and lower is also your demand on the server capacity.

The server makes no decision about what difficulty it sends you. It in fact can't do anything like that. You don't get shares as high as 10M to work. You don't get any shares to work on from the server.

The server gives you some data to work on, and it's a matter of your fortune to find a share. When your client is set to send shares of diff 1, it sends all shares it finds to the server. To find one such basic share you have to compute 2³² hashes at average.

When you set the difficulty to a higher number, only shares of that difficulty (and higher, of course) are sent to the server. This is useful when your share rate would be too (unnecesarily) high with shares of diff 1.

Thanks for the clarification 

Today is a day why most are here on Slush's Pool... 30 blocks found in the last 24 hours.. ( I guess its only 29 as we did have the 1 invalid today)

Well, the situation with the share difficulty is a bit more complicated than I've written so far. There are some pools that set the target (minimum difficulty) for the shares obtained from the client. I've not mentioned them since this thread is about Slush's pool.
You can find those pools in the list Comparison of mining pools at the bitcoin wiki marked as Dynamic in the Variance column.
Once again, this does not apply to Slush's pool.

So, those pools can set different target (minimum difficulty) for the shares accepted from the clients. As your information about the 30 seconds' interval must have come from some of those pools, one could quite well recommend you to look for the answer to your question at the same place.

I could imagine that the client sends shares of diff 1 for some initial time interval in the round (30 seconds mentioned by you), and the server measures (roughly) client's hashrate this way. It (the server) then makes a decision based on this finding, what share difficulty target would be optimal for this particular client. (But this is only my speculation.) This method would have one more advantage of using finer resolution for extra short rounds, where the higher target would cause too a high variance.
[edit]If this "calibration" occurs only after client connection and not in each round, this advantage does not apply, thouhg.[/edit]

Some pools also set the target for shares to a fixed value higher than one. Those pools, however, need no time interval to do so.

What does does a variable diff check mean?
How is that related to the previous discussion to my last post?
Can you post a link that confirmes/explains what you've written?