Relic of the Past  🔗 **Source:** [Relic of the Past](https://share.snipd.com/episode/03879a70-9571-4cf3-a074-afc4aef08630) ## Highlights - **Plugin Submission Volume Has Exploded** - Plugin submissions spiked dramatically, reaching 759 in one week compared to 226 the same week last year. - Luke links the surge to new contributors plus automated tooling and AI speeding reviews, changing repo workload. Transcript: Luke Is it more than that? Last week, we had 759 submissions. That was an all-time record the same the same week one year ago 226 okay so we're almost 350 wow okay so if you go back big bump. Yeah. It is crazy. The amount of plugin submissions we're receiving is crazy, but the plugin team has not only onboarded a few new team members who are really making a difference, but also using a lot of Automated tooling to get through plug-in reviews faster, including AI. Jonathan Now, by your guess, what do you think is the percentage of submissions that are done with AI? Luke Are we talking like 2%, like 3% Why do you keep labeling? No, like 90%, something like that, 80%. Very high. I mean, yeah, I honestly, okay. I enjoy plug and review, Jonathan. Okay, yes. And I'm good at it. Luke, what is it that you enjoy about it before we go ([Time 0:01:01](https://share.snipd.com/snip/33cd5615-5d6c-4f72-8a00-adf6367b577b)) - **Dozens Of Nearly Identical Cookie Plugins Arrive Daily** - The repo is flooded with many nearly identical plugins, especially cookie-consent plugins arriving by the dozen. - Luke describes past practices of spotting tutorial-based clones and blacklisting specific tutorials to stop repeat submissions. Transcript: Luke Of plugins so let's go in that direction because i don't think it's good. Say what you want about AI code. But what we're seeing at the moment is just a flood of plugins all doing the same thing. Okay. Right. Cookie plugins. We're just getting dozens of them every day. How many cookie plugins does the WordPress Plugin Reaper need? Jonathan And are we talking about like recipes for like chocolate chip? No, don't be dumb. Okay, that was too much. Yeah, that was too much. Luke You crossed the line, Jonathan. Jonathan Okay, so that's a great example. There's a problem people are frustrated with, I presume, I'm just guessing here, they're like, oh, I an idea for a better implementation. Or people just following a tutorial. Like you remember back in the early days where it's like, oh, I want to make a plugin. Luke We don't let tutorial plugins on the repo, actually. It used to be back in the day when it was like 100 to 150 plugins per week being submitted. It was a smaller team. And because it was a smaller team, it was much easier to recognize, oh, all of a sudden we're seeing a whole bunch of similar plugins being submitted. And then we go, oh, they all are basically the same code. What's going on here? Oh, it's a tutorial someone's following online. We find the tutorial and we just reply and say, hey, thanks for submitting, but we're not accepting the submissions from this particular tutorial. We've got enough on this. And so we've started talking now about, well, let's keep a list of types of plugins maybe we don't accept, like cookie plugins. ([Time 0:03:43](https://share.snipd.com/snip/49c7da56-4861-457b-9542-1c83d0d1b9b4)) - **AI May Worsen Repo Noise Not Fix It** - Increased submissions are harming plugin discovery and adding noise the repo must manage. - Jonathan predicts the response will be to 'solve it with AI,' which Luke criticizes as recursive and inefficient. Transcript: Luke Like, it's not probably won't read. It's absolutely don't read and just pass straight on to the AI. So, I mean, all right, whatever, Jonathan, I'm over it. Okay. Okay. So with that kind of volume, yeah, go ahead, go ahead. No, I'm just, I'm, how much do people want to listen to a podcast episode where I just complain? You know what I mean though? Like it, think we're running into issues on the repo about, like, finding plugins and plugin discovery. But do you know the most frustrating thing is that the answer is definitely going to be, we'll solve it with AI. That's just how it's going to be. We're going to solve it with AI. We'll find an AI way of fixing it. And that sucks, I think. It's just, like, it's recursive and it's a waste of resources. It's inefficient. ([Time 0:06:01](https://share.snipd.com/snip/4a1adebc-f208-4b44-927b-9e26ce64276e)) - **Repo Is Approving Many Similar Plugins** - Despite noise, many submissions are being approved, so the repo contains hundreds of similar plugins. - Luke confirms the scale by saying the repo currently has dozens of cookie plugins accepted. Transcript: Jonathan Really? So we have hundreds of cookie plugins right now? Luke Yep, absolutely. Okay. That's exciting. Anyway, so here's what I was thinking. Tell me what you think. And actually, Crossword listeners, genuinely reach out. We can find us on our post status. We have a channel on post status Slack. ([Time 0:07:29](https://share.snipd.com/snip/686ff00b-e3ad-4565-af52-09d44cef89e8)) - **Offer An AI Free Manual Review Option** - Offer an AI-free review path where volunteers manually review plugins if authors promise they didn't use AI. - Luke asks if authors would choose a manual, non-AI review option similar to meal preferences. Transcript: Luke What do you think about if there was a way of submitting a plugin to the repo that bypassed any AI checks and was manually reviewed by a volunteer, myself, who didn't use any AI tools in The process? On the condition that you promise you didn't use AI to generate your plugin. Do you think that people would, is that something people want? An AI free path? ([Time 0:08:06](https://share.snipd.com/snip/fb0e593e-3815-4ae5-ac76-9af11c5c6e3e)) - **AI Adoption Is A Choice Not An Inevitable Fate** - Public sentiment outside tech is largely negative toward AI, so adoption is not inevitable; communities can choose to reject it. - Luke compares AI adoption decisions to national choices like rejecting nuclear power. Transcript: Luke You're going to be left behind if you don't have it. I hate that. No, we have agency here. And poll after poll after poll shows that the public do not like AI tools. It's just us in our little tech bubble. It's just Matt Mullenweg who's obsessed with them. And, and we, but we have the agency to say, we reject this technology. We do it all the time, you know, like for better or worse, I think personally, I think for, for worse, we rejected nuclear. Uh, I don't know what it's like in the States, in Australia, we don't have any nuclear power plants, right? That is a decision that we made. And we can make the same thing about AI. And if you look at public sentiment, it kind of seems like we're already going down that track. So don't tell me I'm an old man and I'm just telling people to get off my lawn. No, I'm a progressive. ([Time 0:09:51](https://share.snipd.com/snip/47b1ef1e-48da-4145-a877-fd30a786e35c)) - **Using AI Reduces Your Research And Coding Skills** - Luke views AI as a skill-degrading crutch and says he uses it far less now to avoid cognitive atrophy. - He stopped relying on AI for searches and research to preserve his ability to find and judge sources himself. Transcript: Luke Think of anytime I use AI as taking a loan from the bank. Okay. I'm using AI about half as much as I did four weeks ago. Okay. Because I've had this mindset shift that everything I want to do that AI does for me, it makes me worse at it. Right. So something as simple as searching. I used to use AI a lot to just get me an answer to this query I had. You know, like Australasian, sorry, what is it? There's these two kinds of black birds that hang around near where I live. There's the Australian raven and the Terrasian or something, a crow. And I wanted to be able to tell them apart, which one's which. What does it mean when the eye is white versus when it's black? Is that an indication that it's a juvenile? I could have asked AI that and gotten an answer really quickly, but I'm realizing that my ability to search for this information online is actually deteriorating. I'm not as good or fast at getting these answers as I used to be. And I don't, that's a decision I've made. I don't want to be bad at that. ([Time 0:11:00](https://share.snipd.com/snip/4e3907ce-7ebb-4c13-9075-67ba4c8468c7)) - **ChatGPT Helped But Manual Research Was Better** - Luke tried an academic literature search with ChatGPT and got decent results but later found them inferior to manual searching. - He re-did the search himself and concluded manual research produced better quality citations. Transcript: Luke And I asked ChatGPT, can you do this research for me? And just go and find any academic articles and it has like, you know, the academic mode or whatever and find everything I need in this topic. And I gave it filters and parameters or whatever. And it did a good job actually. And I got this, these results. There was probably 20 usable papers from it that I could combine. I'm doing all the talking on this one, Jonathan. And at the end of it, I thought, wow, ChatGPT is really good at this. I got a great result. But then I just had a second thought. And I mean, this is actually not really my point, but I went and did it manually. And I realized, oh, no, no, chat GPT's result isn't as good as I first thought. It would have been very easy to go, oh, I'm very happy with this. I'm going to run with it. It really wasn't as good as if I did it myself. ([Time 0:12:50](https://share.snipd.com/snip/80043f3f-6b42-4aa1-9858-7a39f67dc7e2))