Hacker News new | past | comments | ask | show | jobs | submit login

No, you don't have an orphaning problem on your own blocks, so it drives centralization.

This is exactly what happened when blocksize jumped from 100k to 250k: every small miner jumped to the largest pool (ghash.io) because they were seeing increased orphan rates, driving that pool over 50% of hashing power.




I don't actually see why that would happen. The chance of a block being an orphan is dependent upon the number of competing miners solving the block at the same time. Why would block size impact that at all?

Larger block size increases computation necessary to solve the block, so it is equivalent to increasing the difficulty. Another factor is that you might decide to wait longer to get more transactions in the block. This reduces your chance of solving the block, but increases your payback. I can't see how it could possibly increase the orphan rate because it increases the possible spread for solution time (some people will wait, while other won't). In other words, it should decrease the orphan rate.

I suppose block size increases the network latency for advertising solved blocks... but going from 100k to 250k every 10 minutes does not strike me as being particularly problematic.

Also, I'm trying to verify your claim and I can't see any evidence of it. It appears that different clients were using different block sizes for a long time. The maximum block size on the BTC network has been over 250K since 2013, according to [0]. Also take a look of this graph of orphaned blocks [1]. It seems to be highly variable between 2014 and 1016, but then lower since then. I see no evidence of a sudden increase in orphans. Clearly, having most of the hashing power in one miner will reduce orphans, but I just don't see a connection at all with blocksize.

So what am I missing?

[0] - http://hashingit.com/analysis/39-the-myth-of-the-megabyte-bi... [1] - https://blockchain.info/charts/n-orphaned-blocks?timespan=al...


> Larger block size increases computation necessary to solve the block

No, they're merkled, so it doesn't.

> I suppose block size increases the network latency for advertising solved blocks... but going from 100k to 250k every 10 minutes does not strike me as being particularly problematic.

Doesn't matter how often it is, I find a block, I mine the child immediately. You have to wait until you've received and validated the block. It's all about latency.

> Also, I'm trying to verify your claim and I can't see any evidence of it. It appears that different clients were using different block sizes for a long time. The maximum block size on the BTC network has been over 250K since 2013, according to [0]. Also take a look of this graph of orphaned blocks [1]

Yep, this is now ancient history. Mean blocksizes tend to mask what's really happening, but I recall the orphaning problem and ghash.io spike personally.

And blockchain.info's orphan block measurements are completely unreliable :(

More background: https://rusty.ozlabs.org/?p=535 (my blog)


Thanks for the reply. I have to say that I still can't quite understand how block sizes affect orphaning, but I'll have to think about it.


That's a fair point.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: