Skipping Along: How Many Is Too Many?

GraVandius

Retired Troll
Pronouns
His Majesty
1572466006175.png

Skipping Along: How Many is too Many?
After the last ERI article I decided to try to look deeper into the skip rates of various recruiters and try to set a benchmark for what exactly should be cause for alarm in the future and to perhaps set a more explicit policy on what to skip in the future. I looked at a total of 18 prominent recruiters and used the same method explained in the previous article, using less than 5 seconds as a benchmark time for everyone rather than guessing their spam timer at the time.. These recruiters account for 302,312 of the 452,245 telegrams sent through the helper, which comes out to a little over 2/3rds of the telegrams sent overall.

1572467517709.png
As you can see above Bri undoubtedly stands out tremendously in comparison to the rest of the data set. Towering over the second place recruiter, which is unfortunately myself. That 20.2% skip rate is the result of not only the natural human error and skipping Nazi/otherwise deplorable nation names but a consistent aggressive skipping of puppets over the past four years. Examining the data closer, I skipped 323 Yuno puppets alone in addition to the regular doses of 5-10 puppets in a row from time to time. Further, 2017 I evidently had the habit of clicking through a few older nations (ones past the window of 20-30 minitues where a manual telegram could reach them) in effort to get a new nation quicker. The combination of those four factors leads to a fairly significant inflation in recruitment totals that is admittedly a bit embarrassing. Overall, I think for the reasons affecting my total at least, the suggestion of an intentional skip button from the previous thread would resolve the issues at hand.

After myself there are 3 other recruiters above the average skip rate of 6.7%, Snowball, Siol Alse and Whitmark. For Siol (The #1 ranked recruiter of all time) and Snowball the cause simply seems to be a higher rate of human error than their compatriots. The skips are fairly spaced out and not in any groups. It's especially understandable for Siol, as the likenesses you get a misclick while recruiting at such a breakneck pace is more common than if you are just recruiting in small batches. Whitmark on the other hand has most of his skips in groups of like 10-15 and are on nation names that do not suggest puppets at all. While it's possible that he was following a similar train of thought to mine in clicking through some old nations he would probably have been under similar pressure to Bri as either Deputy Minister or Minister at the time (March to late May 2019)

On the bottom end of the graph here Prim (1.5%) and Kari’s (0.5%) numbers can be takes as an approximation of what is an appropriate rate for skipping only racist/nazi nations as both profess that is explicitly what they do. Additionally, while CSP has the lowest rate only about 1,000 of his impressive totals were sent through the helper so Drexlore is perhaps our most careful recruiter with a skip rate of 0.2% and 11,000 telegrams sent.


1572465777750.png
The above graph shows the percent of total skips broken down by each recruiter. As you can see, both myself and Siol make up a tremendous amount of the pie as not only do we both have about average skip rates but together have sent roughly 100,000 telegrams. The next largest share is Calvin, who despite a much lower skip rate, when applied to a whopping 40,000 tgs produces a lot of skips. The same can be said for Malashaan, who has over 30,000 tgs.

In conclusion, for the average recruiter I think that skip rates of less that 5% are likely the best as the average of the sample of 6.7% is heavily lifted by myself and Siol for various reasons. Generally it’s important to be careful when recruiting as a misclick results in a nation not getting the option of receiving a much more effective manual telegram. Additionally, if you do mess up you can always go to the activity reports section of NS and show foundings, find the nation and get them their telegram, just like the good old days. Thanks for reading!
 
Another interesting data set. Kudos to you for publishing your numbers even if you were embarrassed. :p
 
This is really interesting, thanks GraV!
 
Mind posting the median since this data is skewed by an obvious outlier?
 
Mind posting the median since this data is skewed by an obvious outlier?
So to clarify the Average Skip Rate in the first chart is the Total # of Skips across the data set divided by the Total # of Telegrams sent not just the mean of the skip rates themselves. The mean by the later method, would be the much higher 8.5% with a median of 2.6%.

Additionally, if you remove Bri's 783 confirmed skips from the Total along with her number of telegrams, the average by my method only falls from 6.7% to 6.4%.
 
I love this - great work, GraV. Goes to show why the ERI is one of our top media outlets!
 
Thanks GraV, neat analysis!
 
This is super interesting! I am glad that this unfortunate incident has sparked some fascinating conversations about manual recruitment.
 
Ah this is lovely Grav! You're definitely a leader for our region in interpreting data!
 
(ones past the window of 20-30 minitues where a manual telegram could reach them)
I will note that while I was an MoI one term, the recruiting tool broke and started consistently giving us nations that were 1-3 hours old, I suspect this wasn't the first time it happened since r3n knew what the issue was and fixed it again, mentioning that he had seen it before.

So, given your TG record, I don't find it surprising that you probably ran into a lot of old nations occasionally.
 
Another interesting data set. Kudos to you for publishing your numbers even if you were embarrassed. :p
I think this is another good reason to think that GraV probably doesn't fall under Article 14 of the CC -- "falsely reporting" -- GraV is being honest and sincere about his, and probably most of our, habits when manually recruiting. I honestly don't think there is any one "skip rate limit" where it suddenly becomes illegal or egregious, it's likely situational in many cases and based on subjective patterns.

GraV skips obvious puppets and perhaps the times of day that he has recruited the most have more puppets in them. It's a highly variable situation and Criminal Code (14)(v) would probably have to be judged on a case by case basis. Even with his higher percentage, GraV seems to fall pretty close to other prolific recruiters, whereas Bri is clearly an outlier.
 
Last edited:
Great analysis, and I think the question of when to skip is a good one. Historically, I've done what it appears others have been doing and skipped objectionable nations and, less often, puppets. Thinking about that now, I'm not sure that actually makes sense though. We almost always have stamp recruiting set (occasionally I don't set a new batch going until a little after the previous batch expired), and we always have a script doing API recruiting. Those methods aren't discerning at all, which means every new nation is going to get a recruiting telegram from us. Manual just increases the likelihood that they will get one sooner (there's also pretty solid evidence that manually sent telegrams have slightly higher success rates, but that's irrelevant here). Skipping nations does not mean they won't get a recruitment telegram from us, it just means they will, on average, get it later. On the other hand, if they get a a manual telegram from us first (meaning before the automated telegrams), the stamped version will be blocked and we won't be charged a stamp for that nation.

In other words, the advantage (such that it is) of skipping is that the nation will likely not see a recruiting telegram from us until a little bit later, whereas the disadvantage is that we re more likely to use a stamp on that nation (which means real money). Thus, we can make the real money that our members contribute to stamps last longer by not skipping. I'm not sure the delay in objectionable or puppet nations receiving is worth the cost of stamp (or even that it's worth anything).
 
In other words, the advantage (such that it is) of skipping is that the nation will likely not see a recruiting telegram from us until a little bit later, whereas the disadvantage is that we re more likely to use a stamp on that nation (which means real money). Thus, we can make the real money that our members contribute to stamps last longer by not skipping. I'm not sure the delay in objectionable or puppet nations receiving is worth the cost of stamp (or even that it's worth anything).
I mean a single stamp is worth .1 cents. To give context, presuming every skip would have beat the API telegram (which is unlikely as for a lot the window in which the telegram could have had likely closed), all the ones included in this study over the last 5 years would have cost us less that $25 total. Manual recruiting barely puts a dent in the sheer shitload of API telegrams sent and definitely provides its main benefit in terms of the increased success rate. Further, in the case of some repugnant nations, if you report and skip them, a decent number will have been deleted prior to the 30 minitue window closing.

Basically, sending manual telegrams to puppets/nations we don't want, is an ineffective use of my time. For example had I sent telegrams to all 323 of those Yuno puppets I would have spent roughly 27 minitues saving the region $0.32 and providing no competitive advantage on bringing new nations into the region. Had I instead picked up more time at my part time job, I could have made about $5.00 after taxes, which is equivalent to 5,000 telegrams in stamps. The benefit in manual recruitment is clearly in the increased success rate, however much that is worth in terms of money is unclear.
 
The amount of time saved is less than that, because it still takes 2-3 seconds to load and skip a nation. I think you're probably right that the amount of money saved isn't worth it though.

I disagree that manual can't/doesn't put a major dent in stamp use. We typically use 6-8k stamps a week on new nations. If we are doing 3k a week of manual, which historically is very doable, that's a significant reduction.
 
We've only hit the 2-3k numbers for a few weeks in the past two years, to my recollection.

I would say we average closer to 300-500 a week more recently.
 
I agree we haven't lately, but that's the point. 2-3k a week is very achievable if it's a focus and that is a significant proportion of the stamp/api numbers.

Edited to add: actually, I just looked at the numbers and we've fairly consistently been around 1k this term. And there was a 3k week in september. We could hit 3k again this week too. A common theme I'm seeing is that recruiting does need work, but it's really not in the dire straits it is perceived to be. The numbers are lower than I'd like but they aren't terrible.
 
Last edited:
We've only hit the 2-3k numbers for a few weeks in the past two years, to my recollection.
This. In my experience the best thing to get individual people sending large amounts, like 700-1,000+, tends to be when there is a "prize" that results in 2, or more, people wanting it.
 
Last edited:
The amount of time saved is less than that, because it still takes 2-3 seconds to load and skip a nation. I think you're probably right that the amount of money saved isn't worth it though.

I disagree that manual can't/doesn't put a major dent in stamp use. We typically use 6-8k stamps a week on new nations. If we are doing 3k a week of manual, which historically is very doable, that's a significant reduction.
Historically though, those 3k of telegrams are not actually saving us any stamps. Due to how a lot of people recruited a huge chunk of those, probably like 50% or more are going to come back as “Previous Recruitment Too Recent” and not save us any money. Id have to obviously look deeper at the data but on a normal “historical” week I’d be surprised if we beat API on more than maybe 1k telegrams.
 
Back
Top