Isaac: Welcome to Never Rewrite. I'm Isaac Askew. Jeffrey Sherman: And I'm Jeffrey Sherman. And today, you can't vibe code a reputation. In in this age, in this age of AI that we are in, where the cost, the barriers to entry are lower than ever, and the cost of spinning up an MVP is lower than ever. Reputation is is emerging as the the bulwark, the thing that i ⁓ becomes the new barrier of. Isaac: Mm. Right to it. Jeffrey Sherman: If I don't know you and I've never heard of you and you're new, ⁓ I trust you less than ever. And so it doesn't matter if you vibe coded something last night and it's amazing, the odds of me trying it are really low. Isaac: Yeah, and I think we can see this too in like the marketplaces for like iOS apps being flooded with vibe coded apps. ⁓ it was pretty flooded. But now it's like hyper flooded. And I I actually I was I was joking with a friend of mine about this about the the workout apps where everyone like w everyone's got a workout well, a lot of people have workout app that they use and they record Jeffrey Sherman: I mean it was already pretty flooded with terribles terrible stuff. Mm-hmm. Mm. Isaac: Like how many reps they did, blah, blah, blah. And then there's some other little thing in there, like, ⁓ if you want to see your history of your workouts, now you to pay. And then I was thinking to myself, ⁓ I'll just, you know, make my own app. That solves my problem. I don't want to pay this person. ⁓ and then like three other people that I was talking to at some different point over the weeks were like, ⁓ I was gonna make my own a workout app. I'm like, Okay, everybody's Jeffrey Sherman: Mm-hmm. Sure. Everybody Isaac: running into this now. And so everyone's just like ⁓ vibe coding their own solution. So now when I go and look at a workout app on the iOS store, I'm like there and there's like there already was hundreds of them. Now I'm like, ⁓ this is all slop. It's the same kind of trust problem I've I've got now whenever I see an article and I start reading it and I see an M-dash or I see a You're not looking at so-and-so, you're doing so-and-so. That kind of wording. And I'm like, ⁓ this is this is chat CPT, isn't it? And so I stop reading. Jeffrey Sherman: Mm-hmm. That that one hurts me because I used I use that ⁓ phrase and I you know if you go look at my blog which has got eight years of history, you will see that I have always used this thing you know before LLMs were even really ⁓ emergent. But i I mean it i you can tell an article it's just the writing is bad. It's it doesn't get to a point, it's general, it's vague, it's wordy, it wanders around. Isaac: Yeah. Mm. Yeah. And it's all the the weird affirmation things like, you're on to something special there. You know, like whenever it responds and you're like, okay. Like like something's off about it. It's like the ⁓ what is that? Like the uncanny valley thing for for visual things for, you know, seeing a face that just seems unnatural. It's like that, but words. And then when you read it, you're like, ⁓ something something's off here. This feels Jeffrey Sherman: Mm. Mm-hmm. Yes. Isaac: Which is like it's almost like ⁓ at when the flood of YouTube influencers happened and now and you pull up your YouTube and there's just a ton of videos. And or in the early days it was just a bunch of random videos, right? And you don't know who has actually got good content versus some kid just making funny noises and laughing and then turns off the camera. Jeffrey Sherman: Mm-hmm. Right. Isaac: You know, it was just random stuff, right? And then o over time, like things started filtering out where like people who were actually good or entertaining or had value kind of floated up. And some people who I don't think have value, but I guess other people do. But Jeffrey Sherman: They they fed them algorithm for whatever value that is. Isaac: Right. But they yeah, somehow it fed the algorithm. and so but then but ⁓ there was one particular area where it was just like a lot of videos could have no value, right? And that's kind of where we're at now with AI, with published articles, with published apps, where now my trust is at an all-time low. You know, is this Jeffrey Sherman: Mm-hmm. Yes. Isaac: A good app, is this a good article? Is this article written by a human even? Or is is anything I'm reading actually real? Or is it just regurgitated stuff? Did someone just go, you know, like a as a j as a when I was in ⁓ college, ⁓ I did brief I briefly was in journalism, and you had to have your sources right for your article. So if something happened, you had to find different sources, live sources, dead sources, and then like make your own version of what's happening of your own retelling with these sources. And now you know, an AI can do that, right? Very easily. Theoretically, I mean I can still get it wrong or hallucinate. But ⁓ it can just go in there and you know easily spit up ⁓ you could ask it, hey, take these Jeffrey Sherman: Mm-hmm. Theoretically. Yeah. Isaac: 15 sources and write fifteen articles about what happened. And it just broop that fast. And you can even say, hey, do it in the style of a sarcastic, entertaining YouTuber. Yeah. Jeffrey Sherman: Right, you could have fifteen news sites and you you could have fifteen different fake reporters, all with the same base sources. Isaac: Yeah. And I actually did this as a thought experiment. I spun up a a website where I had ⁓ like I I I tried to retell it with like a ⁓ somebody from a very stoic fact based reporting versus like an opinion article. And it's very easy to do. Then after I did that, I started kind of feeling sad. Are the Reddit comments that I'm reading on Reddit real, the YouTube comments real. Jeffrey Sherman: Yeah. Isaac: It's like the dead internet theory, right? Jeffrey Sherman: I mean YouTube comments have never been real because they were always I mean it it's always been people spamming the comments for either spam or to reputationally spam reputation, back to the point. One thing that I've noticed emerging is developers having AI write the write, here's the ticket, have AI write the ticket, then have AI do the fix and then be like, here, can you approve this MR? Isaac: Mm. ⁓ could be. Mm-hmm. Mm-hmm. Jeffrey Sherman: And I look at it and I'm like, you you didn't review this. Like it and that is absolutely enraging. ⁓ if you I don't it's not about can AI do the work or is it okay for AI, it's totally fine. Have AI do most of the work, but don't ask a human to review something that you also haven't reviewed because it just shows that like you are do you're disrespecting my time because you ⁓ AI is not perfect. And you have to double check it. I mean, I double check, you know, going back all these years, decades, like before I do a pull request, or even back in the day, I would re I would double check the thing. Like, here's the diffs. I would double check it, and I would often catch goofs in my own work. And I of course I'm awesome ⁓ at my own work. And not doing that when you're Isaac: Okay. Jeffrey Sherman: doing AI and just passing up like, ⁓ well the AI did it. So you know, here, can you approve this? Like, no, I'm not I'm I'm not here to proofread your Isaac: It it's passing off your work, essentially, 'cause like if if you have it generated, ⁓ the the bare minimum you can do is like boot it up locally and make sure like it runs. It solves the problem, like you've QA'd your own work. If you like literally said, Hey, fix the thing, turned your head, and then like said, Okay, it fixed the thing, I pushed it up, can you review it and then like go get launch? You haven't done anything. All you did is prompt. You didn't test any of your work, and then you're asking somebody else to review your work and test it for you. Jeffrey Sherman: Well, th this reminds me, ⁓ you know, decades of experience. I have absolutely worked with developers who would make changes and then ask for a review without running the tests. A and this is before build pipelines were a thing. So we're talking early aughts. They would they there'd be unit tests, but they wouldn't actually have run them and you would run them like you dude, your t unit tests don't work. Isaac: That's kinda low. Right. Jeffrey Sherman: Like this the build is broken. And like, ⁓ Isaac: Yeah. This doesn't even compile. Like, ⁓ it's just a one line change, but they forgot a semicolon. You're like, dude. Jeffrey Sherman: Right. You you've asked me you doesn't compile. And I don't know that developers have gotten any less lazy since then. It's we've added build pipelines and so you know the the CI C D will tell you, ⁓ that y you didn't compile and Isaac: Yeah. Jeffrey Sherman: Yeah, it's just Huh? Like did you did you run it? ⁓ I I I mean I guess that's the whole thing. It's like, did you run it? And so I understand why this mentality is reemerging because it's always been there latently. ⁓ Isaac: Yeah. Huh. Yeah. Jeffrey Sherman: But the with the low quality software, ⁓ and I think we've gone into this a little bit before, where reputation impermanent is more important than ever, like you can't vibe code a mail an email server. I mean you totally can. You can totally vibe code an email server, and even if it's perfect, it won't work because no email provider will talk to it. They will immediately black hole all all the email from it as spam. Because it doesn't have a reputation. Isaac: Right. And I think that's that's kinda like ⁓ and I I kinda somewhat fell into this when when AI first when I started when I first saw the power of AI after a gentic programming earlier this year, and I'm like, ⁓ okay, this is getting better than it was last year. Then I thought, ⁓ SAS is dead. You know, and I think a lot of people did. ⁓ but then I thought, okay, there's some some things like what you're talking about here with reputation deliverability, ⁓ that can't be vibe coded easily or like Jeffrey Sherman: Mm-hmm. Mm-hmm. Isaac: partnerships with like SMS deliverability and Twilio and these other providers where someone's like, ⁓ I can just spin up this ⁓ I don't have to use Active Campaigns Automation Builder. I can just spin up my own automation builder and do that instead. There's still partnerships there. There's still reputation there in in the sense of actual reputation amongst people and reputation amongst like, you know, is this spam. Jeffrey Sherman: Mm-hmm. Right, right. Yeah, if you if you're using a hotmail email address, the odds of anybody looking at your email is very low. I don't know if hotmail's still around. I think they got rid of it, but Isaac: You're right. So there's still I haven't seen the address in a while. I don't know. Maybe. ⁓ either way, yeah. There's some parts of that that still people knowing who you are is going to help you get ahead a bit. Because now it's it's I think it's similar with even applying for a job, right? Where like a lot of times the next job you get is who you know. Like somebody helped get you in the door there. Jeffrey Sherman: I I think Microsoft rolled it into Outlook dot com. But anyway. Mm-hmm. Isaac: Even if you didn't ⁓ apply to work under like for me, I've I've tried to apply at places where there was a manager there that I worked with before, and I'm like, ⁓ I loved working under that person. They were a fantastic manager, right? Or if I have a friend who's at the company who's like, this place is they reached out to me and they're like, This place is awesome. I think you'd love it here. And then like they they talk it up to their people, like, ⁓ you know, Isaac, he'd be a great asset, blah, blah, blah. Right. Now you've got a flood of people who can quote unquote program or even fake responses during the interview or fake or go home and like vibe code the the homework challenge that some people give engineers. The trust there, I don't know because I haven't had the interview in a in a while. But either way, well how you I knew it was a problem because I know you and I were at a conference one time and they were talking about like how do how do you filter through all the people who are Jeffrey Sherman: Mm-hmm. Mm-hmm. Hopefully that's dead. Yeah, I've been interviewed in a long time. Mm-hmm. Isaac: using AI to cheat. And we're like, ⁓ that's a thing There was like a company that's spun up to to to do that, yeah. Yeah. Jeffrey Sherman: Right. And I remember yeah, there were we were at a conference and there were like three or four companies who were all about using AI to as they would put it, leverage your time, but also to reduce the odds of ⁓ the candidates cheating. And ⁓ you and I were like, This is so backwards. Well no it's so bad like if if you are if you are trying to cast a wider net with less effort Isaac: Yeah. This is a problem? Yeah. Mm-hmm. Jeffrey Sherman: And yes, of course you are going to come up with garbage candidates. That's why you like that's part of casting as wide a net as possible. You need to cast a you need to use AI to cast a very small net or a spear. You need to be spear fishing, ⁓ instead of driftnet fishing. Yes. Yes, we we are in the we are deep in metaphors today, friends. Isaac: Yeah. Excellent. I love how I get into these analogies. Jeffrey Sherman: ⁓ and right, because if you're gonna go with AI writing your prompts ⁓ writing your job description, you're gonna cast as wide as net as possible, you're gonna catch people, like at this point especially, y the people you're gonna get are people using AI to apply to every job. And then of course now you need to f to use AI to filter. And AI, you know, it it's most likely gonna filter everyone who is actually unique because they're different. Isaac: Mm-hmm. Jeffrey Sherman: They're gonna fall outside the the params. And so like you're using AI to solve the problem created by AI without actually solving it. Like, but ⁓ you haven't solved any problems. Like you're just back where you started. Isaac: Yep. And I remember I remember they asked like well how do you how do you get good candidates or how do you make sure the candidates that you're talking to are not like cheating? And I'm like, I don't know I just get on call with them and like ask them questions and they're like, Well, what where do you get your questions? I mean I just come up with them. Like I'm I'm just like I'm the like ⁓ you know, here's a here's a problem, how would you solve it? Now we just come up with some random question random problem I've run into before and just talk it through. And they were just like super perplexed by that idea. I guess that because that probably doesn't scale at maybe the corporate level. But for me I didn't have that problem. Jeffrey Sherman: It it what I mean I always my go to is I look at their resume and I ask them to talk about something on their resume. And dear God, I have stumped people. Not even like reaching back twenty years. It's like, ⁓ here's a thing ⁓ that you said you did last or like two years ago. Tell me about that. It's like, ⁓ well I'm like, geez Isaac: Yeah. Right. Yeah. Let's go into it. Using using their own history. Mm-hmm. Jeffrey Sherman: D did you not expect to get asked about the resume? This is this is the point of the resume. I mean, I I've I've been curious about things that are were on resumes that were 20 years old, and the people were like, I don't really remember the details that much. And I believe that. That's fine. And I only asked because I was like, this this thing here, that actually sounds really cool. Can you tell me about that? Like, I don't remember so much. It was it was like this, and like, ⁓ alright, cool. All right, let's talk about something recent. But Isaac: Yeah. Yeah, that's fair. Right. Yeah. Well, yeah, e either way, the that that reputation there and that networking is what kind of gets you in the door, right? Because now we're like even more skeptical than ever that the things we're reading are quality, the things the on the app stores are quality, the person we're talking to is quality or who they say they are, 'cause it becomes easier to kind of fake that. And I I was talking to ⁓ Jeffrey Sherman: Mm-hmm. Isaac: my fiance, Rebecca, about this. ⁓ I w I was asking her about like what kind of problems she's got in her domain. And she's in like a health policy analyst. And so she she starts hitting me with a ton of acronyms and you know, all kinds of things that show that that she has domain expertise in that area. Right? And so when I was thinking, ⁓ what ways can I help her solve problems in her area? I was like, I'm really as an engineer Jeffrey Sherman: Mm-hmm. Mm-hmm. Isaac: I don't know what I'm doing in this area. I'm just like, you guys I I I I need to shadow them to understand their problems. That way I can help with a solution if they even need one. I'm just I'm looking for a problem to solve, right? Whereas from her flow, she's got everything she needs. And I think also too, once Claude becomes more mainstream to non-tech people, and it is becoming that for her too. Jeffrey Sherman: Mm-hmm. Mm-hmm. Right. Isaac: At some point she'll get to the area and I've met one person, one non techie person too, like this, who was like, ⁓ well, I'm just gonna spin up the MVP myself and they did. And they didn't need any engineering knowledge, they just were they're, you know, they knew what they needed and they knew how to prompt well. Jeffrey Sherman: Mm-hmm. Isaac: And so I was ooh, this is an interesting intersection we're at right now, where I can see the people who have domain level expertise. Why do they spend all the time trying to work with an engineer to solve the problem for at least the MVP? 'Cause they're like, I got everything I need here, especially if the engineer c charges money. You know, if it's just like an engineer trying to help, sure, different thing. But if they're charging a ton of money and they go, Well, let me see if I can get something myself booted up before I bring in the big guns, you know, they don't necessarily need that as much anymore, or at least that's where I see the trend. Jeffrey Sherman: Mm-hmm. Right. Isaac: So the actual knowledge. Yes, building your own tools and you know, not needing to spend so much money spending all this time communicating the all the domain expertise they know to someone else or playing telephone game to have it continue down the line. They could they they have everything they need. They're the all the power is now with their domain expertise and the experts, I think. Jeffrey Sherman: Right, people building their own tools. Mm-hmm. Right, and I I think it's an important thing to to mention there. Right. So if if I building a tool to solve my problems and you know, if I wasn't a technical person and Claude gave me something and it was from a technical standpoint garbage, but it it worked for me and my flow and it solved my problem, then it's a great solution. 'Cause it solves my problem. And if it doesn't scale, that's fine. Like and if it's a good if the Isaac: Yeah. Right. Jeffrey Sherman: If the tool turns out to be generally good, then it will get a reputation, because I will share, like I will share it, and other people find it useful, and it will get a reputation. And you know, people will expand it, and that will be good. It's the idea of, ⁓ well, I have this problem and I need this tool, and therefore I'm going to use AI, and it's going to be a generally good tool that it's going to solve lots of people's problems. No, that is not going to happen. Which Isaac: Right. Yeah. Jeffrey Sherman: Is the same that it ever was of you if you want something to go viral, solve a problem and then people won't be able to stop talking about it, right? Virality is a reputation. Isaac: Yes, that's a good point. It's ⁓ it's actually almost like ⁓ anybody who's really trying to solve the problem and is very loud about it, ⁓ it's that's the counterintuitive thing is like, ⁓ look at this thing that solves everyone's problems and everyone's like right. The more you try to like try to push your own solution, especially with like the the word used there like general solution, in in this era where everything is very niche and everyone's like a custom coded solution because it's easier to do that now. Jeffrey Sherman: Mm-hmm. Isaac: than I ever was. That those that actually solve the problem, they get talked about for you. You don't have to advertise it. Right? If it actually solves the problem. And I remember this was like there was a similar similar thing with thing we talked about in prior episodes with working and shadowing the operations team. And like I found out, ⁓ they're doing all this manual stuff. Jeffrey Sherman: Mm-hmm. Right. Isaac: And now that I'm here and I g and I see what they're doing, I can use at the time this is before AI, I can I can write some code that will help solve their problem. And I think we're in a really great era now where there's the ability to ⁓ 'cause you know, like every company's got their own like internal tools and their own internal admin dashboard where they can do this or that. But they're all very different by company 'cause they're custom little things coded to solve tiny little problems here. ARC. Jeffrey Sherman: Mm-hmm. Mm-hmm. Isaac: ⁓ the tools and arc. Yeah. Yeah. Jeffrey Sherman: Well and they're all emergent. Right. So you can have a conversation about, ⁓ well, what's your control plane like? And people will know what you're talking about. Well, hopefully people know what you're talking about, but like everyone's got their own. ⁓ until somebody writes a book and says, Okay, these are the this is how these are the things that you need for like for observability, right? There was a long time where yes, everyone ha needed observability and people had metrics and people had logs, and then somebody came up with ⁓ open telemetry. Isaac: Mm-hmm. And ⁓ yeah, yeah. Yep. Mm-hmm. Jeffrey Sherman: Like it emerged. And it's like, okay, now open telemetry has emerged. It's has this reputation. You could not have vibe coded the open telemetry spec or or the concepts 'cause nobody would have bought in. 'Cause they're it didn't have it it emer like reputation emerges. You can't it does not you cannot have reputation emerge like Athena from Zeus's head. You you you you have to Yeah, right, w I I'm through I forgot the axe part where Isaac: Mm-hmm. Yeah. Nice. Okay. ⁓ I love when we get into this. You cannot vibe code reputation, yeah. Jeffrey Sherman: Yeah you anyway ⁓ yes we're getting long too. ⁓ Yes you cannot Vive code reputation I think is bringing it back. And so putting in the work Isaac: Mm. Jeffrey Sherman: Solving problems, right? Outcomes over everything. Isaac: Solving the problem solving real problems that people really experience versus ones that you assume or hope or predict they will experience, ⁓ will always be be preferred and and will speak for you. All right. Any last comments about v vibe coding reputation? Jeffrey Sherman: ⁓ no. ⁓ well this podcast started off with no listeners and now we have dozens even. ⁓ as we've earned that reputation. ⁓ I we we'll have to look ⁓ we did not vive code the people. W but we could have. We could have hired bots to inflate our numbers. Isaac: Dozens. We didn't vibe code the people. The listeners. And it would have fallen off as soon as the money stopped flowing to the hired bots. Yep. Jeffrey Sherman: And it would have fallen off like it wouldn't have gotten us anything. It would have been entirely vanity metric of Yeah, pain. But so yes, don't hire vot bots to inflate your metrics that they're not real. Also, it's all about reputation, real reputation. Awesome. All right. Thank you all for listening. I'm Jeffrey Sherman. Isaac: I'm Isaac Askew, and this is Never Rewrite.