Programming Diary #19: Follower network strength and preparing for Open Source publication

Summary

Once again, I didn't have much progress to report at the two-week mark, so I skipped that update and it has been about a month since Programming Diary #18: WhaleAlert inspired activity notices. For this interval, my activities mostly shifted from the Steemometer programming to alignment of voting weights with the strength of an author's follower network. I also made some progress towards getting the Steemometer ready to publish as open source.

Background

Here are the two goals that I set for myself in Programming Diary #18: WhaleAlert inspired activity notices.

1. Implement algorithm changes in automated voting so that voting percentages start incorporating information about an author's follower network;
2. Start preparing the existing Steemometer code for Open Source release by adding comments, removing clutter, and trying to make sure that it doesn't leak details about my development environment.
   

I expect to revisit the first goal many times in order to continually improve the voting alignment between blockchain content and the human attention that the content receives. I also expect the second goal to take more than 2 weeks of effort.

I made progress on both goals during the last couple of weeks, but as expected, I have work remaining before I can call either one complete.

Progress

Aligning voting weight with follower network strength

For the first goal, aligning voting weight in my autovoter with the strength of an author's follower network, here's a visualization of the rules that I implemented. (each line is based on the number of followers that an author has)

Here are some characteristics of the "Follower Score" as implemented:

1. If an author's follower network has a median follower reputation of 25 or less, it doesn't matter how many followers they have. The follower score is 0.01 (to avoid multiplying by 0).
2. If an author has more than about 375 followers, and the median reputation is above 25, then the median reputation doesn't matter. They get the maximum follower score, which is 1.
3. In between 1 follower and 375 followers, the follower strength increases from 0.01 to 1 with the number of followers and with the median follower reputation.
4. To get the follower score above 0.5, the author must have somewhere between 64 and 128 followers.

I don't really like the shape of these curves, but it's a first pass. I ran into trouble with API responsiveness for a few days, so I'm still observing the impact of the changes.

As previously mentioned, once I find a curve that I'm happy with, I'll use that to inform the vote suggestions in the Steemometer tool.

Preparing Steemometer for Open Source publication

In order to do this, I am leaning about javadoc, and how it is implemented in Netbeans. I also learned about the SLOC maven plugin, so that I can contextualize my progress through the various Java class files. At present, it seems that my project currently contains about 2,790 lines of code.

Package Name	File Name	Type	Blank	JavaDoc	Comment	Code	Total
	module-info.java	src	0	0	2	9	11
steemometer	ApiServer.java	src	7	29	1	52	89
steemometer	AuthorInfo.java	src	44	61	22	133	260
steemometer	CountOps.java	src	34	41	34	227	336
steemometer	MedianCalculator.java	src	20	39	1	108	168
steemometer	PostDisplayManager.java	src	15	43	3	91	152
steemometer	PostInfo.java	src	42	122	8	227	399
steemometer	StartPoll.java	src	6	10	0	21	37
steemometer	SteemData.java	src	45	112	11	192	360
steemometer	SteemPriceFetcher.java	src	21	9	12	65	107
steemometer	Steemometer.java	src	190	12	121	1078	1401
steemometer	SteemometerConfig.java	src	14	41	0	50	105
steemometer	SteemometerController.java	src	5	10	1	23	39
steemometer	SystemInfo.java	src	4	12	0	9	25
steemometer	TransactionNotifier.java	src	30	60	14	217	321
steemometer	UrlValidator.java	src	16	44	1	68	129
steemometer	VoteSuggestor.java	src	5	17	0	15	37
steemometer	XferMemoInfo.java	src	23	93	1	126	243
steemometer	XferMemoManager.java	src	15	42	3	79	139
2 package(s)	19 file(s)	java	536	797	235	2790	4358

I learned how to use Netbeans to insert javadoc comments in all files, and I have reviewed the auto-inserted comments in the ApiServer.java file and the AuthorInfo.java file. I also went through all files and removed auto-generated comments that I thought were unnecessary.

Next up

Since my goals from the last cycle are still in progress, I plan to continue them for the next two weeks.

1. Experimentation and improvement towards aligning voting weights with follower network strength.
2. Getting Steemometer ready for Open Source publication.

As with last time, I don't really expect to finish either goal in the next two weeks.

Reflections

I want to follow-up on three topics from the Reflections section in Programming Diary #18: WhaleAlert inspired activity notices, and also add a new topic:

SBD print threshold / Steem inflation

The Steem price has now remained above the SBD print threshold for a little more than a month, so I have continued to keep an eye on Steem inflation. The first quarter was deflationary, but the first week of the second quarter reversed that trend. It will be interesting to see what happens in coming weeks since the BTC halving is expected on April 20.

Blockchain activity

In the last post, I wondered if SBD printing would lead to an increase in blockchain activity. It appears that the answer is "yes". In that post, the blockchain activity was reported at about 185 operations per minute. Now it's up to about 207.

Steemit.com stability/API performance

The overall stability problems that were mentioned in Programming Diary #18, have been solved. API reliability still has occasional problems, however. As noted above, my autovoter fell behind by about two days for some reason. I didn't notice it for a while, so I have no idea what the cause might have been. Also, simple python activities are often still blocked by the rate limits that were implemented last fall.

Witnesses

Here's the new topic for today's post. I have some thoughts about witnesses in the Steem environment:

1. The top-20 witnesses have more to do than just validate blocks. The task of setting the price feed seems to be in good order, but they also need to consider the SBD interest parameter. I've discussed this topic here, here, and here. It would be nice if the witnesses would weigh in with their perspectives. Is there some reason why 0 is the optimal percentage?
2. I could be wrong, but I think there's at least one top-20 witness who has disengaged considerably from the ecosystem. It might be good for people to reevaluate their votes.
3. Since I don't control enough stake to influence the ranks of the top-20, I often focus my witness voting further down in the ranks. There are a number of disabled witnesses who are still in front of real witnesses, so - again - it might be good for people to reevaluate their votes. For those of us who have smaller stakes, moving votes that aren't needed from the top-20 section to higher ranked witnesses might have a substantial impact towards downranking the disabled witnesses.
4. I've mentioned it before, but IMO, @moecki should be ranked much higher than #58. It seems to me that he provides more to the ecosystem than several of the witnesses in top-20 positions, and many of the others, too. I hope readers will consider allocating a vote towards his witness.

Conclusion

If slow and steady wins the race, then I'm on track for a big programming win. 😉 It seems like publishing the toy Steemometer program as Open Source is in sight now, even if it's still not imminent.

My purpose for breaking out of iterating new features for it was to force myself to develop some other skills that will be useful for eventually implementing a peer to peer social media client. Based on the experiences learning about javadoc and SLOC, that seems to be proceeding as hoped.

I hope to see you in two weeks with Programming Diary #20!

---

Thank you for your time and attention.

As a general rule, I up-vote comments that demonstrate "proof of reading".

---

Steve Palmer is an IT professional with three decades of professional experience in data communications and information systems. He holds a bachelor's degree in mathematics, a master's degree in computer science, and a master's degree in information systems and technology management. He has been awarded 3 US patents.

---

^{Pixabay license, source}

Reminder

Visit the /promoted page and #burnsteem25 to support the inflation-fighters who are helping to enable decentralized regulation of Steem token supply growth.