WEBVTT

00:00:00.380 --> 00:00:04.120
Speaker 0: Do you still read the code or should use to read?

00:00:04.320 --> 00:00:04.780
Speaker 0: The Code?

00:00:04.900 --> 00:00:06.820
Speaker 0: that's a question at discussion.

00:00:06.900 --> 00:00:15.880
Speaker 0: That came up on tech X over the weekend In my bubble, at least and it's an interesting question obviously also nice rage bait Question.

00:00:15.940 --> 00:00:20.600
Speaker 0: but that's not what this episode shall be about And I'll answer the question.

00:00:20.760 --> 00:00:25.000
Speaker 0: It's not as simple yes Or no for me and i also want to hear your thoughts are really do.

00:00:25.180 --> 00:00:26.500
Speaker 0: so please share them too!

00:00:27.900 --> 00:00:32.040
Speaker 0: and I'll get back to this question here, but i think another question has come first.

00:00:32.800 --> 00:00:35.080
Speaker 0: Do you still write the code?

00:00:35.180 --> 00:00:40.020
Speaker 0: Because obviously if-if you still right all your code... You will read it!

00:00:40.400 --> 00:00:43.160
Speaker 0: Right so yeah..you don't need to read because you wrote it.

00:00:43.340 --> 00:00:54.920
Speaker 0: But for many developers myself included We probably DON'T write ALL our code And you can be anywhere on a range of zero to one hundred percent.

00:00:55.540 --> 00:00:58.920
Speaker 0: Regarding that question how much code you still write?

00:00:59.120 --> 00:01:10.220
Speaker 0: I probably am pretty close to a hundred percent by now already and That definitely changed a lot over the last six eight months as it did for many developers.

00:01:10.360 --> 00:01:18.260
Speaker 0: I'm sure with AI models getting better, but most importantly those models being Better at following instructions.

00:01:18.320 --> 00:01:22.920
Speaker 0: Being fine-tuned for following instructions better and then the surrounding tools like Claude code and so on.

00:01:23.000 --> 00:01:43.420
Speaker 0: So that changed for me And I've talked about that in another episode how that also sucks a lot of joy out of that craft and I'm trying to find new joy in the building stuff, but this is off course.

00:01:43.620 --> 00:01:45.060
Speaker 0: the first important question.

00:01:45.240 --> 00:01:46.260
Speaker 0: Do you still write?

00:01:55.600 --> 00:02:12.700
Speaker 0: I think it's too early to tell what the right place is on that scale, because we're still early in that entire AI evolution when it comes to coding and will have see long-term effects of more software being fully AI generated as.

00:02:20.020 --> 00:02:21.620
Speaker 0: But what about that other one now?

00:02:21.760 --> 00:02:22.700
Speaker 0: Do you still read the code.

00:02:22.760 --> 00:02:26.260
Speaker 0: Well, again this is complex because we can split it up.

00:02:26.740 --> 00:02:35.920
Speaker 0: We have reading the code and I would say we have caring about the code And for me That Is not The same.

00:02:36.020 --> 00:02:46.960
Speaker 0: and You Can also Read but Not Care Which Is kind of Dumb and i'll get Back to that or not read and care which is Possible i Would argue and It has Implications.

00:02:47.640 --> 00:02:51.160
Speaker 0: so as Mentioned you may not read and not care.

00:02:52.060 --> 00:03:01.780
Speaker 0: You may READ AND NOT CARE, you may NOT READ BUT CARE... And of course you may REED AND CARE!

00:03:02.500 --> 00:03:11.360
Speaker 0: These are the four combinations if you have this split-off of reading & caring Now If YOU DON'T READ and you don't care.

00:03:11.800 --> 00:03:13.860
Speaker 0: that is my definition of vibe coding.

00:03:14.280 --> 00:03:20.020
Speaker 0: Now you may have a different definition and that is fine, but this is my definition of vibe coding.

00:03:20.280 --> 00:03:21.580
Speaker 0: Not reading not caring.

00:03:21.640 --> 00:03:22.560
Speaker 0: That's why I'm quoting now.

00:03:22.680 --> 00:03:25.160
Speaker 0: that does NOT mean that you don't care about the product.

00:03:25.300 --> 00:03:31.060
Speaker 0: You probably do because if you DON'T CARE ABOUT THE PRODUCT WHY ARE YOU DOING SOMETHING IN THE FIRST PLACE?

00:03:31.140 --> 00:03:33.580
Speaker 0: RIGHT BUT YOU MAY JUST CARE ABOUT THE PRODOCT.

00:03:33.640 --> 00:03:34.600
Speaker 0: IF IT WORKS YOU'RE HAPPY.

00:03:34.700 --> 00:03:37.980
Speaker 0: IF THERE IS AN ISSUE YOU JUST TELL THE AI HEY BUBBLES NOT WORKING.

00:03:38.040 --> 00:03:41.800
Speaker 0: PLEASE FIX AND YOU DON'T CARE ABOUT THE CODE AND YOU DON'T READ THE COTE.

00:03:42.160 --> 00:03:43.960
Speaker 0: that is my definition of vibe coding.

00:03:44.800 --> 00:03:49.480
Speaker 0: Now, if you read the code but don't care... That's kind of weird!

00:03:49.780 --> 00:03:51.980
Speaker 0: I guess your just killing time in this case?

00:03:52.200 --> 00:04:00.480
Speaker 0: If u reed but not care.. You may be at a job where you're paid to do something so spend your time analyzing codes when you dont really care.

00:04:00.540 --> 00:04:05.180
Speaker 0: i dont know Not really my focus here But sure ,you could have that combination.

00:04:06.120 --> 00:04:11.420
Speaker 0: Here are combinations which are more interesting though Not reading and caring and reading and caring.

00:04:11.540 --> 00:04:13.100
Speaker 0: Now let's start with reading and carrying.

00:04:13.180 --> 00:04:19.860
Speaker 0: I guess that's the obvious one, i guess you could call this AI powered software engineering or something like that?

00:04:20.399 --> 00:04:35.100
Speaker 0: um because you may not write the code anymore or not all of it but you care about it as a software engineer you maybe responsible for it and you read it because since you care About It Since You May Be Responsible For It Well...you want to see what is going on there right?

00:04:35.160 --> 00:04:35.660
Speaker 0: Makes sense.

00:04:36.300 --> 00:04:37.600
Speaker 0: So that's an obvious one.

00:04:37.760 --> 00:04:40.720
Speaker 0: And I'll get back to it, but what about not reading and caring?

00:04:40.800 --> 00:04:41.380
Speaker 0: Is this even

00:04:44.700 --> 00:04:45.160
Speaker 0: possible?".

00:04:51.160 --> 00:04:52.440
Speaker 0: Worrying answer first.

00:04:52.720 --> 00:04:57.920
Speaker 0: I don't know obviously, I don' t know if that is the future Just as four years ago?

00:04:58.320 --> 00:05:02.880
Speaker 0: I wouldn't have imagined That one day we would not be writing to code anymore.

00:05:03.760 --> 00:05:06.040
Speaker 0: I may just not be imagining yet that one.

00:05:06.080 --> 00:05:08.620
Speaker 0: they were Not reading the code any more at all.

00:05:09.580 --> 00:05:15.040
Speaker 0: In reality though We would probably end up on a scale like this and i Would argue we already are On a scale Like This.

00:05:15.140 --> 00:05:16.480
Speaker 0: but again i'll get back To this.

00:05:16.780 --> 00:05:18.240
Speaker 0: so not Reading And Caring.

00:05:18.680 --> 00:05:20.040
Speaker 0: Is This Even Possible?

00:05:21.060 --> 00:05:27.040
Speaker 0: As I mentioned, i would say yeah it probably is but its a totally different way of thinking about software.

00:05:27.600 --> 00:05:49.560
Speaker 0: It relies on you trusting the AI not just when it comes to writing code but your also using AI for reviewing and scanning or analyzing your codebase for auditing your code base Fable five or whatever, or codex.

00:05:49.680 --> 00:06:03.440
Speaker 0: Or pie my favorite agent with your favorite AI model to work on the code and then you could use The same agent in a new session With the same or different model?

00:06:12.460 --> 00:06:18.480
Speaker 0: You can and you probably should also spend a lot of time planning.

00:06:18.720 --> 00:06:29.200
Speaker 0: And reading those plans, reading those specs... I think not reading the code if done right- If that is the future!

00:06:29.940 --> 00:06:32.320
Speaker 0: ...should NOT imply that you don't read anything.

00:06:32.600 --> 00:06:47.540
Speaker 0: You SHOULD absolutely spent more time then creating good plans to gather with AI making sure Working on the right stuff with the right Implementation details if you want to put it like this.

00:06:47.620 --> 00:06:51.220
Speaker 0: so in the right way You don't just care about the product.

00:06:56.280 --> 00:07:02.000
Speaker 0: Ensure, or not ensure but increase the chances of getting good results there.

00:07:02.460 --> 00:07:07.920
Speaker 0: if you spend more time on the specs On the plans and then let AI implement it.

00:07:08.160 --> 00:07:15.340
Speaker 0: And You Then have code reviews where you give their reviewing agent also access to that spec Or two that plan the AI should implement.

00:07:16.000 --> 00:07:21.580
Speaker 0: now still Of course one problem you'll Have here which is why I'm NOT ON THAT.

00:07:21.940 --> 00:07:29.480
Speaker 0: i don't read anything train that you don't know with certainty what's going on in your code base.

00:07:29.540 --> 00:07:32.920
Speaker 0: You can't, just can't know if you're not reading it!

00:07:39.040 --> 00:07:57.480
Speaker 0: and you don't read it at all, And You just sign It.

00:07:57.600 --> 00:07:58.800
Speaker 0: You Don't have certainty.

00:07:58.980 --> 00:07:59.840
Speaker 0: What's in there?

00:08:00.360 --> 00:08:02.040
Speaker 0: And you may or May not be doing that?

00:08:02.520 --> 00:08:05.980
Speaker 0: I Personally don't do That.

00:08:06.400 --> 00:08:09.440
Speaker 0: i personally Do Still Read Code but!

00:08:10.060 --> 00:08:12.580
Speaker 0: Thats Now the Interesting Part.

00:08:12.660 --> 00:08:15.780
Speaker 0: Maybe Again On A Scale.

00:08:15.820 --> 00:08:19.480
Speaker 0: I DON'T READ ONE HUNDRED PERCENT OF THE CODE.

00:08:20.320 --> 00:08:23.940
Speaker 0: I Have a Hard Time Telling You Where Exactly I am on That Scale.

00:08:24.920 --> 00:08:36.400
Speaker 0: It may be somewhere in the middle because I review all code AI gives me, but of course that does not mean they read every single line necessarily.

00:08:43.659 --> 00:08:51.600
Speaker 0: are the important ones.

00:08:52.120 --> 00:08:59.760
Speaker 0: I dive into those, analyze them and then dive in to some parts where i know that AI likes to mess up!

00:09:00.240 --> 00:09:07.080
Speaker 0: That is how our review and i may just scan or briefly skim over the less-important part.

00:09:07.800 --> 00:09:18.000
Speaker 0: For example when it comes to the AI writing code for reading environment variables I may not read every line there.

00:09:24.500 --> 00:09:46.020
Speaker 0: I don't know how this will change.

00:09:46.080 --> 00:09:54.140
Speaker 0: How it may be moving closer to not reading one hundred percent now, i don't Know if i'll ever get two One hundred percent?

00:09:54.260 --> 00:10:10.920
Speaker 0: i can't look into the future and i guess If we get to one hundred for both writing and reading We really have see what that means for us as software engineers, it may still be super important to give AI the right tasks.

00:10:11.040 --> 00:10:15.420
Speaker 0: To know about designing systems and about architecture... And this would.

00:10:15.600 --> 00:10:20.940
Speaker 0: my guess is that we are moving in this direction but obviously that's hard to tell!

00:10:22.060 --> 00:10:27.540
Speaker 0: The question will be if at some point you wouldn't write any code or read any code.

00:10:28.120 --> 00:10:35.240
Speaker 0: If you need to know How to code right if it's worth being able To read or write code, right?

00:10:35.280 --> 00:10:37.180
Speaker 0: If that still important and Right now.

00:10:37.280 --> 00:10:44.320
Speaker 0: It definitely still is for me And definitely still Is also because the question is what do you care about?

00:10:44.540 --> 00:10:44.700
Speaker 0: right?

00:10:44.760 --> 00:10:48.060
Speaker 0: That's The other part caring reading as one Part.

00:10:48.140 --> 00:10:52.860
Speaker 0: but we read Because We care and we may Care Without Reading As I made it clear.

00:10:53.160 --> 00:10:54.160
Speaker 0: But What Does Care?

00:10:54.300 --> 00:11:00.120
Speaker 0: Even Mean and i would say care Obviously Can mean a lot of things.

00:11:03.620 --> 00:11:26.720
Speaker 0: It may mean something like the code style, you make care about the code-style how functions are created which approaches use their... You may even care about bigger decisions like if you want object oriented programming functional programing stuff like that um..You may care About the readability of your program!

00:11:30.500 --> 00:11:32.660
Speaker 0: and about stuff like reliability.

00:11:33.180 --> 00:11:39.760
Speaker 0: And there are many other dimensions we could think of, but There are some dimensions here which I say matter less now.

00:11:39.940 --> 00:11:45.700
Speaker 0: code style readability especially if you don't read not too important it's just important.

00:11:45.940 --> 00:11:50.820
Speaker 0: in that case i mean the AI is able to Read and understand the code.

00:11:50.860 --> 00:11:52.960
Speaker 0: And it most likely will be fine with that.

00:11:53.580 --> 00:11:58.100
Speaker 0: now Code style can still matter also for token efficiency reasons.

00:11:58.520 --> 00:12:03.160
Speaker 0: if you have a code base where You just have a lot of unnecessary checks which as mentioned is something AI right?

00:12:03.260 --> 00:12:11.880
Speaker 0: Now tends to do or some AI models tend to That will also be extra tokens to burn when the AI analyzes that code.

00:12:11.940 --> 00:12:16.960
Speaker 0: So, that is still some thing here but it may not matter as much.

00:12:17.500 --> 00:12:24.200
Speaker 0: in the past I mean we all had our own preferences regarding code style and for me that definitely changed.

00:12:26.340 --> 00:12:34.640
Speaker 0: Now obviously though stuff like security reliability testability That still matters a lot.

00:12:34.860 --> 00:12:43.180
Speaker 0: Obviously, that is the part if anything in software you want it to do what should be done.

00:12:43.820 --> 00:12:51.140
Speaker 0: You may care about performance and should not crash randomly on your users.

00:12:51.700 --> 00:12:54.400
Speaker 0: And of course these are all things we're seeing with modern softwares.

00:12:54.600 --> 00:13:13.600
Speaker 0: Modern software often isn't reliable but may NOT be secure Not achievable, especially with these AI models being quite good at finding security vulnerabilities and us not necessarily having access to the latest model that can do but you should still care about it.

00:13:13.700 --> 00:13:14.480
Speaker 0: That is my point!

00:13:14.820 --> 00:13:18.900
Speaker 0: You should try to build secure reliable performance software.

00:13:19.280 --> 00:13:30.500
Speaker 0: in my opinion And as a software engineer, that of course also means you may want to give your input regarding which libraries to use in our project.

00:13:30.560 --> 00:13:46.220
Speaker 0: So this is where we then move more into the systems design or software architecture area and these are things.

00:13:48.320 --> 00:13:52.500
Speaker 0: Having good specs, having good plans is important.

00:13:52.820 --> 00:13:55.360
Speaker 0: if you get closer to one hundred percent not reading.

00:13:55.800 --> 00:14:02.260
Speaker 0: Which again may not be the case for you This is all something where there is no wrong or right as I try to make it clear.

00:14:03.220 --> 00:14:07.440
Speaker 0: so These parts do get more important specs, plans.

00:14:08.080 --> 00:14:12.060
Speaker 0: Because again you want to definitely ensure security and reliability.

00:14:12.500 --> 00:14:18.000
Speaker 0: if you don't care about code style or readability as much anymore which may very well be the case.

00:14:19.020 --> 00:14:22.640
Speaker 0: And therefore of course this entire question off Do You Still Read?

00:14:22.700 --> 00:14:33.780
Speaker 0: The Code is obviously Also a bit of rage bait as I mentioned before and not as easy to answer And there just is no right or wrong here.

00:14:33.940 --> 00:14:36.100
Speaker 0: And by the way, this should be obvious.

00:14:36.180 --> 00:14:45.740
Speaker 0: but There also is a big difference if you have Something some software running in production maybe being used by thousands hundreds of thousands Or millions of people?

00:14:46.300 --> 00:14:49.700
Speaker 0: And it all depends on which area are your working in.

00:14:49.840 --> 00:15:00.200
Speaker 0: i mean If you're working in banking let's say Errors in your software may have catastrophic Consequences and not just banking.

00:15:00.260 --> 00:15:05.580
Speaker 0: There are even more important parts of banking obviously health defense tech all that stuff.

00:15:05.680 --> 00:15:09.260
Speaker 0: so That is an important aspect.

00:15:10.080 --> 00:15:16.680
Speaker 0: If you are an indie hacker with zero users, yes sure You may not care about the code too much.

00:15:18.340 --> 00:15:32.440
Speaker 0: But overall I think we can still see a trend or maybe just as with writing code, where all of us are moving closer to one hundred percent for not reading the code.

00:15:44.080 --> 00:15:45.960
Speaker 0: Can we care without reading?

00:15:46.120 --> 00:15:48.420
Speaker 0: Will that really, really work.

00:15:49.020 --> 00:15:55.540
Speaker 0: I'm not sure for me... ...I don't have the trust in AI and i am not sure if or when it will change.

00:15:56.540 --> 00:15:57.220
Speaker 0: It may change!

00:15:57.460 --> 00:16:11.620
Speaker 0: Not ruling out just as writing code has changed But right now im still more into read-and-care part Or somewhere between because As mentioned Im somewhere in middle of that scale.

00:16:11.840 --> 00:16:14.840
Speaker 0: But as I also mentioned, please let me know where you are.

00:16:15.160 --> 00:16:19.060
Speaker 0: Zero percent reading one hundred percent reading maybe still a hundred percent writing.

00:16:19.160 --> 00:16:21.380
Speaker 0: that may also be the right way to go.

00:16:22.060 --> 00:16:27.880
Speaker 0: Let me now and let me your thoughts about this entire topic in general