Quote from: envy2010 on February 03, 2014, 08:40:09 PM

112/100 = 12% change up, 100/112 = 10.7% change down. This has been noted on previous pages.

The difficulty dropped below 1 because even if 35 blocks were solved INSTANTLY after the fork, the 36 block average time would still have been over 10 minutes -- and so even after 35 drops of 12% the algo "thought" the diff was still too high. Though it wasn't instant, once a couple blocks were found the next 30 or so came very fast. Then, very suddenly, those long blocks were far enough back to be left out of the average, and the average time dropped to below a minute.

The 36SMA12LMT system is intended to react fast to lots of hash entering and leaving the network (which it does well) but a side effect of that is that it overreacted to what it perceived as a lot of hash having "just left" a the time of the fork. If the difficulty had started at 50, block times would have been more reasonable for the 36 before the fork, and this oscillation would never have started.

Also, having the hashrate rise faster than it drops is good in that it prevents hyperinflation. As long as the drop back down is fast enough to keep block times reasonable (at least under an hour), that's more suitable for reaching an equilibrium state.

Check out the rest of my comment, You've answered me before I wrotte the second part, also thank you for 10.7 information not alot of us had this piece of information as we all got the 36SMA12LMT as 12% goes both ways.

Quote from: zerodrama on February 03, 2014, 08:54:10 PM

Quote from: kuroman on February 03, 2014, 07:36:11 AM

Quote from: zerodrama on February 03, 2014, 03:51:50 AM

We are a community coin. We don't add features that most people can't relate to. If we add black magic, fewer people will contribute to the vitality of the coin. Then we die.

We are going to restructure the coin so that this stuff is easier to deal with.

Correct me if I'm wrong, but from your comment I get the feeling that you didn't make the effort to look it up and understand how it work, and I think it's fair to make a judgement about it after you do those things. and btw it is a simpler solution than SMA36LIM12

You are even wronger. SMA36LIM12 is simpler than the ArbitrarySciFiVariableName spaghetti you posted.

It's not whether it works. Maybe it's perfect (except for WTF broken FPU!), but it matters if the average person understands it. If they don't, then they have no voice.

Quote from: SlimePuppy on February 03, 2014, 08:50:37 PM

Kuroman, I like the analysis you've provided in the past. I understand what you're saying about our cycles here and think I can help build a bridge between you and Mav. It's important, first, to remember that this fork was not intended to be a 100% solution. It could have been with a bit more time (or less wasted time dealing with trolls) but we didn't have the time available. This is designed to get us moving and to make sure we never got stuck too high or too low again. As Mav has said multiple times in various venues, this is at best an 80% solution. We CAN stay here for months and ride the waves - we'll be alive, we'll be attracting enough miners, and end-users can actually start using the coin. This gives us a working platform the PR team can work from - and they're kicking ass!

Yes - we can and will do better! And now, because of this fork, we have plenty of time to create a 100% solution without having a gun to our heads.

A quick additional comment - we know from our testnet work that this algo WILL converge and stabilize. We didn't have that before. Wink

Also - keep in mind that in conjunction with this fork we also got a significant endorsement from the Oprah Winfrey empire - I'll guarantee you that we didn't see that coming. Cheesy

I hope you and envy will continue to work towards the long-term goal of getting CAT to Mars. Cool

Andy

Andy thank you for your comment, the fact is the 12% limite is supposed to make the diff converge indeed, but that is under some strict conditions such as, fix hashrate jumps, the jump of hash rate is made when the block time is 600s ....ect which is definitly not the case not to mention that the SMA36 makes everything goes beserk and due to other variables it is countering the convergence from the 12% limite.

Quote

It uses this function KGW = 1 + (0.7084 * pow((double(PastBlocksMass)/double(144)), -1.228))
So if fast rate is high the function above is applied if/when it's the opposite 1/KGW is applied
Here is the whole code behind it, there is nothing magical about it just a simple math formula applied dynamicly :

No explanation of the function. Not magical. Magical. I don't think it means what you think it means. So, at this point it is magical.

I have looked at it. I think I know what it does. I also think it can be done a lot better. Without magic.

I over-estimated your understanding capabilities it seems, if you can't figure out as much that you have KGM applied when hashrate increase drastically and 1/KGW applied when it deacrease dramatically while KGW is a clear formula and the code it self is clear than writing the SMA36LIM12.

If you say it's hard then explane to me why every single new scrypt altcoin use this solution? You are just being overprotective for a solution that doesn't work properly.

Quote from: zerodrama on February 03, 2014, 09:06:20 PM

Kimoto is not simple.
Kimoto antagonizes the coders against the users.
Kimoto is not explained.
Kimoto has competition.

Let me fix that for you

-You don't get it =/= not simple
-You don't know how to code (neither do I since it's not my domaine of expertise but I can read C and can understand alghorithms) but since almost every new scrypt altcoin uses it, it proves how wrong your are and how it is easy to implement.
-It is explaned and I posted the code which is simple, so again if you don't get it and having trouble understanding does mean it is not explaned, and I'll give you an example, even if I give a primary school kid the book or just the lesson on a high school math subject, he will not understand it and think it's antagonizing, not explained... ect ect he will have the same reaction as you right now, while the lesson on it self is nothing much just some high school math.
-And what the problem if it having competition (why don't give some example while we are at it) and competition is good, it means that the best solution will prevail, and will push developers to innovate more, Quantum Mechanics woudn't be where it is if it wasn't to it competition with classical physics and general relativity, if it wasn't for the competition between Nicolas Tesla and Thomas Edison, you won't be using alternative current at your home and electronics won't have involved as much... ... ...

Quote from: SlimePuppy on February 03, 2014, 09:55:20 PM

This may be obvious to you, kuroman, but how many people on this forum do you think are at least marginal C programmers?

Let's all use words, examples, and conversations that are available to the entire community. Limiting the conversation to a few, either through expediency or hubris, doesn't help the greater good.

No solutions are off the table right now - not all possible solutions are ON the table yet. LOL We're not yet at the stage where we need to fight over options. Thankfully, we now have time to propose and more importantly TEST the solutions on the testnet, not just on an Excel spreadsheet or in a Matlab pageprint.

Everyone - put your 'thinking caps' on and get those ideas on the table!

Thank you!

I don't understand your point here? I posted the KGW code to people that are concerned about it and were claiming it is overly complicated which is not the case as proven is explaned several times.

The solution works, not on a testnet which doesn't represent the real life conditions, but on real coins that've been and being hit by profitability pools.

NB: this is why I didn't want to join the catcoin-dev channel discussion despite being present and folowing what happening there, people are blankly dismissing other people ideas, I saw some heated discussions, where no one want to make compromises. Also calling people math geek/nerds and discrediting them on that same channel doesn't help really because and let me remind you however you break the discussion the basic level of all this will be math and code.
And this kind of agressivness and stance that pushs people off (especially from Zerodrama), and it is the reason why I stoped and many did at some point when the fork plan was initiated from commenting, and just to make things clear, I'm not criticizing the fact that we've forked, and I understand very much the principale on which the solution works on and how it was supposed to solve the problem heck I even defended it against people that made a rushed judgement and said it doesn't work at day one of fork and told them to let trough some cycles my comment is still there, but after 9-10 cycles I thinks it's fair enough to have an initial judgement and I've given my analysis on why it doesn't work, and a fix from my percpective while rooting for an available solution that is proven to works properly .