Progress Report for Calibrae: Universal binary build in pprocess and initial wiki page up with most basic changes described and specified
I
have not got a huge amount done this morning, but I do have @enzan now working on creating the universal 64 bit x86 binary build script.
The build script is fairly simple, but takes a bit of time to do. It starts with creating a chroot environment based on Debian 7, which is as old as is practical. The necessary prerequisites, GCC 5, and Boost 1.60 need to be either built or possibly binary versions installed in the chroot, and then the actual cmake script is executed.
Next, we use a tool called AppImage, which turns the binary into a fully self contained binary file that will run on any version of linux (the kernel) from the version in Debian 7, or later. It will run within a Docker container directly, also, within Windows or MacOS.
This is the goal of all this - to make it so users can just download a binary, and run it.
After that, I will be, or someone will be, writing a simple program, preferably one that can also be compiled to a universal binary using AppImage, that automates the creation of the configuration. It could even be written in javascript and work from a form on a web browser.
Initial Wiki for calibrd now available
https://github.com/calibrae-project/calibrd/wiki
This wiki contains the majority of the proposed changes, and importantly, the new concept that came to me as I was doing a Q&A with the Whaleshares Discord folk. I realised that Mute and Follow could alter parameters that can have a suppressive effect on silent accounts, which is how vote bots work. I realised there had to be some kind of formula that scales the effect of reputation of follow/mute accounts upon their targets, and I decided for an initial concept it would be an inverse square function.
The formula that effects rep based on mutes and follows sounds interesting. I look forward to seeing it in action!
so upvoting this post and commenting on it would have used up 2 of my transactions.. I am interesting in seeing if I can adapt to calibrae
Yes, we will be testing it before we roll it out to get the numbers right. 24 seems like a reasonable amount to start with, but maybe it should be 32 or 48.
I am going to start counting ..the upvote I gave you and this comment count as 2 .. see how many I have already done just on your post lol..sounds fun to try .. :)
It's a great exercise, I think. When we get it running in the testnet we will need a lot of people with different ways of interacting, and we will figure out what is typical and what is the range.
It isn't gonna be a hard, simple limitation, think of a fuel tank, that fills up to a certain amount, and when you use it, it depletes it. The tank can fill up to a certain amount, if you do nothing for a week, or so, and then it can't fill any further. It will take a little tuning to get it right, but once we know what is reasonable for humans, we can use this as a way to stop bots and spammers, because their behaviour is outside human parameters.
Bear in mind I'm a very casual Steemit user and I don't concern myself with what actually happens under the hood when I click stuff... Could Calibrae allow users to 'refill' their tank from their own spendable funds if they go through their share of the reward pool too fast? I know users can always send transactions but that's a clumsy interface compared to a simple upvote.
I have been thinking about it more, and there is actually a bandwidth system hidden inside Steemit, but I think it is too generous, I have never seen it low enough to stop my activity. It should, really, and I should see it. Have a look at this:
Don't you think that bar should be in the interface at the top to tell me how much I have left? You can see, because I have a small account, that I have consumed quite a bit of my bandwidth. I am looking at this again, and I think it simply needs to be exposed, and some proper user testing done to see what real humans actually need in terms of bandwidth, because bots can use far more than this.
What about using a tiny python script (using bottle) to make an easy to use browser based configuration tool? The user would just point their browser at the localhost:8000 or whatever, and it would take/validate their choices to make the config file(s), and maybe compile/run it.
That would work too.
@berniesanders again downvoted you with his bots. Just can't wait to see the new calibrae platform where I hope I will not bump in such people that are willing to destroy anything that's not in line with their interests.
I tried to post all of dostoevski's 'crime and punishment' on his idiot, $165 buck worth post, of about 150 words. It got tiring after a while. Took me a couple of hours. A lot more valuable labor than he would have any understanding of.
I proved something: he can't vote me down endlessly. He stops and I am at zero. Then he thinks, with his -16, that he proved a point. Can't have me joining me down the bottom. How about I flag myself also! yes!
Why 0 rep?
flagged by a whale.
Please if anything, do not allow accounts to smash other accounts out of spite. I understand the need for checks and balances like up voting and down voting, but they should have equally proportional power. If a post or comment gets 4 up votes and 1 down vote, the down vote should only have a 20% effect since it is 1 out of 5. Something like that.
I have already devised a scheme for this. It's central to the changes, in fact:
https://github.com/calibrae-project/calibrd/wiki#reputation-coefficient
The exact details of the implementation may change a little, and I haven't yet fully studied the way reputation works as it is but the principles are as follows:
Calibrae will leverage the ability of humans to recognise sybils, scammers, trolls and scammers, and allow the community to judge them by suppressing their ability to operate their account, both through downvotes, as well as being able to effectively downvote the account using mute.
I have thought quite a ways beyond that. You, reasonably, do not think that stake based power is reasonable. Well, neither do I, alone. This is why we are putting reputation on an equal footing with stake. The individual vote is not going to help balance things in a situation with potential sockpuppet voting and bot voting. The point of stake regulating vote power is to put a risk for the bad actor. This is nullified by the fact that there is a small group of bad actors on this platform who have stake that cost them nearly nothing, and zero risk because they can just suppress their enemies with downvotes.
If you want to learn more about calibrae, come join the forum: http://calibrae.freeforums.net/ and you can also voice your opinions about what you learn. It is a work in progress and we haven't moved far beyond specifying the parameters of the system due to the aggressive suppression of my attempts to inform people about it. You will find that the plans are very reasonable and balanced, and you can ask questions to further clarify. The more voices speak, the more questions we hear, the more we can adjust our system to what it should be.
We do not want to repeat ANY mistakes here, no matter how well intentioned they were.