Decentralized alternatives like Mastodon and Lemmy are gaining popularity in response to growing complaints about centralized social media platforms such as Twitter and Reddit, which frequently prioritize business interests over user experience. This study employs a mixed-methods approach to scrutinize the impact of Reddit users’ migration on Lemmy. It elucidates user growth patterns, revealing significant registration spikes and centralization trends within decentralized structures. A sentiment analysis with VADER, incorporating a dataset of 48,272 comments from before and after the migration, depicts a predominantly positive sentiment towards Lemmy and criticisms of Reddit. A comprehensive survey with 354 responses from major Lemmy communities validates and supplements the findings, shedding light on users’ motivations, adaptation experiences, and long-term intentions. Furthermore, qualitative interviews with 16 purposively sampled users offer in-depth insights into individual experiences, community dynamics, and perspectives on decentralization and engagement. This study reveals a promising future for Lemmy, highlighting its adaptability and users’ commitment, and contributes valuable insights to the discourse on the sustainability and growth of decentralized platforms in a dynamic digital landscape.
I think that the metrics that the author is looking at, while easily accessible, are the wrong metrics to be considering on impact.
First, we need a way of quantifying the quality of users, rather than quantity. Specifically, we know that no ‘apparent’ material dent was made in the user numbers, and actually probably wasn’t. But that doesn’t mean that critical players in the system didn’t hit the eject button. I think its a widely held belief that it was more of the ‘power’ user types that left and went to lemmy. We need to test that specific belief. This work doesn’t do that.
There are however metrics that could test that belief. For example, posts/ user and comments/ user both can be representative of engagement. Log interactions/ log system interactions can be used to identify power users (like @thepicardmaneuver). This can also be done at the community level (we’ll have to do some matching), to compare individual communities.
I think also a classification of “S-tier” engaged users, “A-teir” engaged users, users, and bots; calculating a ratio of them, then plotting them against log engagement is likely the best way to describe ‘overall’ engagement. My hypothesis is that Lemmy has a significantly higher ratio of S and A tier, while Reddit has a higher ratio of lurking users who never posts (shadow accounts) or comment, and bots. The ratio of these factors together create a metric of ‘quality’, which is a dynamic expression related to quantity ~ engagement. To determine quality, we need to know how engaged are users with the community, and how often.
I’ve been working separately from this different approach but am obviously not close to publishing. I wasn’t even necessarily planning for peer review outside of the lemmy community. I was thinking more of a bot to gather and present these metrics at regular intervals.
Are you the author @Blaze?
I think that the metrics that the author is looking at, while easily accessible, are the wrong metrics to be considering on impact.
First, we need a way of quantifying the quality of users, rather than quantity. Specifically, we know that no ‘apparent’ material dent was made in the user numbers, and actually probably wasn’t. But that doesn’t mean that critical players in the system didn’t hit the eject button. I think its a widely held belief that it was more of the ‘power’ user types that left and went to lemmy. We need to test that specific belief. This work doesn’t do that.
There are however metrics that could test that belief. For example, posts/ user and comments/ user both can be representative of engagement. Log interactions/ log system interactions can be used to identify power users (like @thepicardmaneuver). This can also be done at the community level (we’ll have to do some matching), to compare individual communities.
I think also a classification of “S-tier” engaged users, “A-teir” engaged users, users, and bots; calculating a ratio of them, then plotting them against log engagement is likely the best way to describe ‘overall’ engagement. My hypothesis is that Lemmy has a significantly higher ratio of S and A tier, while Reddit has a higher ratio of lurking users who never posts (shadow accounts) or comment, and bots. The ratio of these factors together create a metric of ‘quality’, which is a dynamic expression related to quantity ~ engagement. To determine quality, we need to know how engaged are users with the community, and how often.
I’ve been working separately from this different approach but am obviously not close to publishing. I wasn’t even necessarily planning for peer review outside of the lemmy community. I was thinking more of a bot to gather and present these metrics at regular intervals.
I’m not, I barely crossoposted this from nutomic’s post on !lemmy@lemmy.ml !
Hey buddy I see you everywhere now. Just want to say I appreciate your participation on Lemmy.
If you click on “Author Information & Copyright” there is a contact email for the author.
No. I’m not.