Ecommurz Tweet Analysis: between Layoff and Data Breach

Iqbal Rahmadhan
8 min readOct 2, 2022

As a human being and a tech employee, getting updates on what is going on in the community is necessary. Especially in the past few months where we have much hot news, from the saga of a gov ministry and data breach to sad news of massive layoffs around tech companies.

A cat in Minecraft named Ecommurz (image by author)

In this social media age, you can get this information easily from many sources, but not all of them can give in a short time. Let me introduce Ecommurz (or this account doesn’t need any introduction), a famous cat account that occasionally shares these kinds of news and becomes a source of truth for many people.

Even though we know the basis of this account is mostly on Instagram, but its tweets on Twitter also attract people to give a reaction. We will dig out tweets from Ecommurz to understand what happened recently and how they can bring people’s attention.

Ecommurz on Twitter (screenshot by author)

Objective and Method

The goal of this analysis is I want to know what is the topic that engages people the most in the past few weeks by looking at Murz’s tweet’s public metrics.

I worked with Twitter API v2 and Tweepy library, and I heard from many people about the advantage of using them to get the data. Do not expect any advanced method here (like topic modeling or sentiment analysis), maybe we will cover it sometime in the next post. Initially, I am quite amazed that with this API, we can gather the context of each tweet using Context Annotation, but unfortunately, many tweets have no value in it, so I decided not to use this feature yet.

Back to the goal, to achieve that, here are some specifications that I use to gather the dataset:

  • Look for 100 newest tweets, with the latest date being September 23, 2022.
  • Exclude retweets and replies.
  • Extract public metrics of each tweet including the number of retweet, quote retweet, reply, and like.
  • Each tweet has an original ID but it consists of a long number. To simplify this, we generate a new shorter ID.

With the above specifications, the dataset will look like this sample where we can find that the earliest tweet that has been captured is on July 30, 2022.

Dataset sample (image by author)

Okay, let’s get the analysis started!

Tweet Analysis

Question 1: How are Ecommurz tweets’ characteristics?

First of all, from one hundred tweets, we will look at the distribution of the engagements. Since we only can access the public metrics from the API, in this story I will use the number of retweet, quote retweet, reply and like to represent the engagements on each tweets.

Okay, let’s see how is the distribution looks like with the box plot.

Box plot showing the distribution of public metrics on each tweet. First chart shows all the public metrics, second chart shows public metrics without number of like. (Image by author)

By looking at the first chart above, it is clear we can see that number of like significantly tend to be bigger than other metrics. The consequence is the other metrics will have less contribution if we calculate engagement as the total of these four public metrics. If we remove the like numbers, as you can see in the second chart, we can get more clarity on how is the contribution of each other metrics.

Question 2: Is this behavior also observed for other popular accounts?

Before we take any action, let us do a comparison with other popular accounts to see if this behavior is also observed on them.

First, we take the other popular “meme account”, Elon Musk. The second one is HRD Bacot, with consideration because these two accounts are talking about a work-related issue. The last one is Extra Time ID, one of the football accounts that I followed.

Let’s plot the distribution of each account’s public metrics:

As we expected, the number of like on these four accounts also dominate the other metrics with different scale. Same with Ecommurz, let’s see if we can get some insight if we remove like numbers.

Exactly, we can get a clearer picture here and realize there are three types of behavior.

First is Ecommurz where the quote and reply numbers are relatively the same and below the retweet number. In this case, people tend to share the tweet without the need for any additional comment (quote) or discussion of it (reply).

The second one is the behavior of Elon Musk and HRD Bacot where retweet has the biggest range but has the median that is not much different from the reply. It can be interpreted that people tend to share the tweet but also discuss the tweet in the reply section.

The last one is the interesting one and I am pretty sure of my assumption. Football accounts like Extra Time ID usually share about football reports and mostly troll some football clubs. This behavior can trigger people to come and start the war in the reply section, making Extra Time ID get the number of replies that tend to be bigger than retweets and quotes.

By comparing the behavior of these four account, we get the conclusion that we need to remove the number of like to get a better sense of how people reacts to a tweet.

Question 3: What is the topic of the tweet with the highest number of engagements?

Next, let’s sort our dataset based on the total number of retweets, quote, and reply so we can focus on the top fifteen tweets. Additionally, we also categorize each tweet manually. Here are the top fifteen of one hundred tweets.

Top 15 tweets based on number of retweet, quote, and like. (Image by author)

Interesting that we have two major categories here. The first one is about layoffs that happening in several companies like Shopee (tweet ID: 20), Garena (46), Zenius (90), and the most recent one Indosat (0). Besides the layoff story, we combine it with the general downturn of the startup that was mostly reported from Shopee and SEA (49, 40, 25, 67).

The second topic is about one of the ministries in the Indonesian government that people love the most. Their new regulation about electronic system providers (PSE) triggers the creation of Gatotkaca (84), the masterpiece that soon will replace Google. And of course, the saga about the data breach and their response (45, 43, 52, 50, 34).
Sadly that we have these two topics dominating our top fifteen tweets, but among them, there is one good news appears Tiket got the highest visit/minute record when they sold Seventeen tickets (78).

Now we will take a deeper look by plotting each tweet in relation charts of quote vs retweet, reply vs retweet, and reply vs quote. Please note that in the following chart we only highlighted the top fifteen tweets with the above two major topics. The first topic (“Layoff Story/Startup Downturn”) will be plotted with orange color, the second topic (“Kominfo/Data Breach”) will get a blue color, and other than that will get grey color. Besides categories, we also make the size of each point represent the number of likes. So the larger the point, the bigger the number of likes.

As we can see on the above scatter plots, our top three tweets are consistently appear in the top-right quadrant. This means that these three tweets always get highest attention by Murz followers by retweeting, quoting, or replying the tweet.

Interestingly enough that tweet number 45 a joke response to news about Kominfo asking a hacker to please not attack has a number of replies and quotes that are significantly higher than the other two tweets. My assumption is instead of only retweeting these tweets, people also tend to give comments and more jokes.

After this funny tweet, Murz also reported how the hacker responded to Kominfo’s statement in another tweet. And guess what, that tweet is tweet number 43 also in our top three list. After these two tweets, we have the highest number in total, the tweet about when the Shopee layoff happened.

Top three tweets screenshot (Screenshot by Author)

The other tweet that we should take a look at closely is tweet number 78. From the plot, we can observe that this tweet depends on the number of quotes to come up to the surface. With this curiosity, I try to look at the history of quote retweets of this particular tweet but unfortunately, we cannot open the quote with the page showing “No Quote Tweets yet”. To know what happened, I proxy the quote retweet with replies and turn out the replies mostly talk about the struggle and how the app often crashes.

Conclusion

We already completed the analysis and got some conclusions:

  1. By comparing with other popular accounts, we understand that number of like always tend to be the largest among public metrics, and the characteristics of each account can be analyzed by looking at the number of the retweet, quote retweet, and reply.
  2. For Ecommurz, the top fifteen tweets that sorted by the combination of three metrics show there are two main topics that attract people to react:
    - Data breach issue
    - Layoff story and other startup downturn topics

It is so unfortunate that layoffs become one of the highest attention topics in the tech world. For those who have been impacted by this situation, Ecommurz (and DesignRant) created a form to help you connect with employers. Please go and check here: https://murzfeed-app.com/

Also, you can reach many recruiters of Glints (I believe they are everywhere on LinkedIn haha) or just go to https://glints.com/ to find out a bunch of job openings.

Good luck to all of us and of course, happy birthday Ecommurz!

Let’s connect with me in LinkedIn.

--

--