Shadowing Practice: Episode 531: Shallow feedback and my coworker is a huge sloperator - Learn English Speaking with Video

Choose a practice mode

Each mode opens a full-screen practice room.

583 sentences
If sentences are too short or too long, click Edit to adjust them.
1
it takes more than becoming a subversion version control hipster to
2
be a great software engineer this is episode 531 of the
3
soft skills engineering podcast where i'm your host jameson dance i'm
4
your host dave smith soft skills engineering is a weekly advice show about all of the non -technical things
5
that go into the technical field of software development like
6
going back to the old ways to the the tools
7
that really lasted you know like subversion what was before subversion was
8
that like cvs yeah cvs i never used
9
that one i just heard about how much better subversion was than it yeah
10
and cvs was built on rcs
11
and of course microsoft had their own thing called visual source save
12
that was parallel to both of them
13
and it was horrible huh well just imagine how much better your life would be
14
if if all your problems caused by git weren't there
15
and you had a whole separate set of problems yeah totally different all shiny
16
and stuff for a while yeah dave that's not what our show's about it's about the magic of
17
other stuff like thanking our patrons do you want to do
18
that yeah big shout out to those
19
that are supporting us at a level where they get a weekly shout out
20
and the very first name i don't know how to pronounce
21
oh yeah i i had to fumble through this live last
22
time they are yure chukalev plan your next trip with trip
23
with www .wayla .travel actually I joined to fund Jameson's yacht oh Bucharestuscarawas Pataskala Kinnikinnik oh boy hi
24
Pika, Ganananhutin Skioto, Geuga, oh alright, Willwood what Lattice Washerum Seth, Jenny Kim, the Stokasparit Ira Chan,
25
Jonathan King, all one word to show you love AI, put an upside down pineapple on your cubicle
26
or office door who even knows what their actual names are
27
my actual name on linkedin is yammy debugging in the dark
28
candy token limit reached jacob channeling will angel of death how
29
do you write a joke how do you write a joke in 50 characters by
30
that was funny will corner angel no javier gonzalez space chewy open print salted fries semicolon last episode 100
31
jncc oh there's an emoji with a flag has the most soft skills listeners per capita quick what country is
32
that iceland all right jncc iceland flag nice uh we have
33
next up won't angel dan from drone deploy never is not
34
just a crater on mars flamingo emoji i like chicken i
35
like liver please deliver i can see dodds kyle boss you are doing much better than you think Kent C.
36
Dodds And we've got an awesome Set of emojis here that spell out Soft skills Emojineering
37
Does that start Surprised that made it through Patreon Yeah, it did really good It started out with a skillet The skillet emoji,
38
which is the International symbol for fans of the podcast The soft skillet emoji Oh, I didn't even notice that,
39
yeah You must be in dark mode too I am, yes.
40
Yeah, this is very well done.
41
I just see it everywhere, too, with all the fans we have.
42
Right, just unending skillets.
43
Just blends in.
44
Yeah.
45
Soft skills, emojineering.
46
Nicely done.
47
Very nice.
48
That is just art.
49
Okay, Jenny Kim, The Stochastic Parrot, Ira Chan, Jonathan King, Zanai, Beautiful Functional User Documentation, Adrian at Kappa .ai,
50
open paren, empty name to protest the new name like Limit, Will Angel.
51
Will Angel has relapsed and is playing Factorio.
52
The letter R.
53
Oh, no. Are we losing this patron?
54
We're about to lose Ragnar.
55
Ragnar's going away one letter at a time.
56
Oh, man.
57
Brayden Cain's John Grant, Brittany Ellick, closed parenthesis, followed by a literal closed parenthesis, closed parenthesis, followed by a literal closed parenthesis.
58
Oh, my gosh.
59
That one kills me.
60
Man, I hope we don't lose Ragnar.
61
That would be really sad.
62
I've also noticed that Will Angel has not updated their Patreon name in a while, and I feel like that means the Factorio relapse could be pretty bad.
63
Yeah, this is bad.
64
It was not bad enough to say that he had relapsed But it is so bad that he hasn't updated his
65
We better do a Wellness check
66
Dave should I read our first question Let's do it Okay
67
this is from a listener who claims Their name is Jamie ain't No dance
68
Jamie ain't no dance instead of James is I think so Hey space therapists Yep that is one of our titles.
69
One of our credentials.
70
I'm a consultant that helps big companies to fix all kinds of stuff.
71
It's ridiculous how many different hats I wear.
72
One day I'm the interim PM.
73
The other day I am telling SAP devs how to use a coding agent.
74
Next day I gather stakeholder requirements for a customer self -service portal.
75
I do all these tasks on a fairly shallow level
76
and all of that work is happening without much coordination from my company.
77
I think I do good work, but it's hard gathering feedback on my work since no colleagues are involved, and I can't simply ask the customer without revealing insecurity.
78
Send sophisticated tips and don't make me quit.
79
Cheers.
80
We've been undermined so many times by listeners who know the shtick.
81
Well, 531 episodes.
82
Time to get a new one.
83
Yeah.
84
Keep your job.
85
Quit your shtick and get a new one.
86
There you go, sir.
87
That's our mission quit your stick
88
so my interpretation is this is their colleagues they're talking about are like the consulting company
89
but they they're kind of solo from their company at this
90
other the at these other clients right that's what i got
91
to hard guarding feedback on my work i don't know maybe this is a cultural thing
92
but i feel like if i paid some consultant especially
93
if they were kind of this jack of all trades many hat wearing
94
if they're just kind of doing a fixed job but
95
as a consultant instead of a full -time person that feels a little bit different.
96
But if they're more vaguely focused on help me solve problems really well, I think it would make me feel good if they said, I really want to do a good job
97
and I want to know how you think I'm doing because I would like to wow you with how awesome it's going.
98
Tell me what you think is going well.
99
Tell me what you think you'd like to be different.
100
Well, you've probably seen, Jameson, those bumper stickers that go on fleet vehicles that say like, how's my driving?
101
Call this number.
102
Oh, yeah.
103
Yeah, maybe just put one of those on your laptop.
104
Yeah.
105
I saw another bumper sticker that said, I have to poop.
106
That's why I'm driving so fast.
107
And my kids like that one a lot.
108
Oh, I bet they loved it.
109
Your kids and me.
110
Yes.
111
So put that one also on your laptop.
112
I have to poop.
113
Next to it.
114
That's why I'm typing so fast.
115
That's why I want to end this meeting early.
116
Yes.
117
Yeah, I don't know what i can't simply ask the customer
118
without revealing insecurity no yeah that's false i think this person feels insecure
119
and then projects
120
that uh okay space sorry space therapist right i mean this
121
is this is like we are credentialed to do this
122
so you have to take this advice yeah you you brought
123
it up yeah yeah it's like it's like breaking hearts in hearts where okay you said it
124
so now yeah you do space therapy that's right so here's my space therapy you are you feel insecure.
125
And so you then ascribe this response onto people that they're just not going to have when you ask them for feedback.
126
Now, I mean, it's one thing to be like, am I doing a good job?
127
Please tell me.
128
You know, it's kind of like very different.
129
But if it's another thing just to say, hey, you know, as part of client engagement, when I finish an engagement, I like to gather feedback so we can make sure we're delivering the best possible service for you.
130
Yeah.
131
And here's like a three -question survey, and two of them are just click on the number of stars, you know, and then it's like the third one is just write a sentence
132
or two about how you felt about the service you received
133
yeah yeah we work with some consultants for specialized things not kind of general problem fixers
134
but they send that stuff out to us too yeah probably
135
and sometimes we ignore it and sometimes we actually do fill it out yeah exactly
136
and that's the other thing is if someone doesn't give you the response don't assume
137
that it's bad also like every time you get service from
138
some kind of it help you know you're gonna get one
139
of these things anyway right like are those people insecure i would be
140
that job would make me insane i would crumble that's like did you turn it off
141
and back on again uh please leave me a five -star review yeah
142
and then they're gonna leave you a one -star review
143
because the internet was down right there was nothing you could
144
my cellular data connection could only send one star maybe there could be some sensitivity around this
145
If you are working with people that you feel like aren't pumped, you're there.
146
Yeah.
147
Maybe someone's brought you in, but the actual people you work directly with are a little grumpy.
148
Maybe, yeah.
149
Feels like, well, our boss doesn't think we can do our jobs, so they brought this outsider in or something.
150
Yeah, to tell us how to use a coding agent.
151
Yeah, but surely someone there has decided to bring you in and is paying you.
152
You should be able to ask them.
153
Yeah, their opinion probably matters the most.
154
Yeah.
155
Like there's some executive who said, well, this team needs to level up their PM and coding agent use and stakeholder requirements gathering.
156
And so I know they're not up to it.
157
That's why we brought you in.
158
And yeah, they hopefully have the good sense to know that the people are grumpy is not feedback necessarily.
159
Although they might not have good sense. And so.
160
That's true.
161
Well, no, they brought you in.
162
Oh, well.
163
They do have good sense.
164
Ah, yes.
165
You do run a risk, you know, that you're going to get some negative feedback.
166
And I don't know, I guess you might be inclined to hide that from your own boss or something.
167
I don't know.
168
I'm not saying you should, but I'm saying that that might be something you worry about.
169
But, you know, I was thinking there might be an opportunity to do something bigger here.
170
If you're a consultant that helps big companies, I wonder if your company already has some kind of mechanism in place for this that you don't even know about.
171
Yeah.
172
Yeah, that's what I was actually just thinking that.
173
Yeah.
174
Can you see any of that?
175
Do they send it out to their clients?
176
Exactly.
177
And if they don't, I'd be frankly kind of surprised.
178
Yeah.
179
It's always a little fraught
180
because are you doing a good job often gets reduced down to does whoever gave the feedback really like you or not?
181
And those are sometimes the same thing, but not always.
182
Yeah.
183
But yeah, I would be surprised
184
if the company was just sending you out they're billing you at some rate
185
and they have no idea just has no idea yeah they just find out
186
when the checks stop coming like oh i guess crap i
187
guess they were unhappy huh yeah shoot what can you do
188
well i guess we just don't get the money yep well that's not usually how businesses operate
189
so yeah i mean i i'd say tap into that um
190
and don't be afraid to ask for feedback and
191
if you formalize it
192
and just make it seem like it's part of your standard operating procedure it won't come across as insecure.
193
I would really hate to work somewhere where me asking questions and exploring things and saying, well, how do you think I did on this would make me seem insecure because,
194
or cause me to lose face or I don't know.
195
I would hate that to be so on my guard
196
that I can't reveal any hint of doubt or weighing things or questioning or wanting to get better.
197
How could I possibly ask for feedback when I'm already definitely perfect?
198
Yeah.
199
Yeah.
200
Feedback implies I'm not already doing the best possible job.
201
Right.
202
I can't allow that.
203
Which is impossible for me to even consider.
204
Yeah.
205
I think they would appreciate it.
206
Now, not if you do it every day or every week, but maybe at the middle and end of each engagement.
207
I'm assuming these engagements last like a few weeks or a month.
208
Although this person did say one day I'm a PM, the next day I'm SAP devs.
209
If you're literally only spending one day with people, I don't know, how much feedback can you really get?
210
How much legitimate understanding does your client even have?
211
What you're hoping for is to tap into the folklore
212
that is now spreading about you across this company of rare sighting of the Sasquatch.
213
Like, oh, I saw that consultant.
214
Yeah, they just appeared out of nowhere and told us about Claude Code and just disappeared into the mist.
215
And we still think about them to this day
216
and then you kind of scoop those up together into a new folklore
217
and then that's your feedback what kind of legends do they tell about you exactly
218
and how long do those legends persist yes that's the true feedback yeah although they probably remember longer
219
if you're just outrageously rude and scream at them a lot
220
somebody I don't know they were just nice and pleasant
221
and understood our problem said oh here's what you do
222
and then bailed that won't stick in their mind as much so you have to adopt some personas and some eccentric habits.
223
Oh, yeah, totally.
224
Do something truly surprising.
225
Smoke two cigarettes at the same time.
226
Oh, outrageous.
227
Outrageous.
228
That'll help.
229
Have we answered the question?
230
I think so.
231
I'm looking forward to the legends of the SAB consultant who smokes two cigarettes simultaneously.
232
Yes.
233
Yeah.
234
What a story they'll tell.
235
Yes.
236
I think they just told it.
237
They smoked two cigarettes at the same time.
238
It's kind of a short story.
239
Yeah.
240
All right.
241
Dave, do you want to read our next question?
242
Let's do it.
243
This comes from Tired of Load -Bearing Vocabulary.
244
If any of you use AI regularly, you already know what this question is going to be about.
245
All right.
246
The question asker says, can I tell a coworker to stop using Claude?
247
I work on a small team at a mid -sized company, and a coworker keeps dropping overly verbose pull requests that are all full of AI jargon.
248
These are increasingly hard to understand.
249
And then when I ask for design intent clarification, I often get sent a long document that is even harder to understand.
250
Oh, man.
251
I hate this so much already.
252
We haven't even gotten to the punchline.
253
Okay, continuing.
254
And the code sucks.
255
Misleading doc strings, overly complicated functions, and weird places where the AI has convinced them that something needs to be backwards.
256
And oftentimes, I end up having to fix failing tests
257
that my coworker probably didn't run my co -worker probably didn't run
258
or report back later that their changes have degraded something in prod
259
or is causing a lot of logging noise i worry
260
that i'm doing the same thing
261
and just not noticing code is starting to pile up in review
262
and i don't want to be rude
263
and comment ai slop on 50 of the code what can
264
i do yeah the the world is changing this is a problem
265
that could not exist at this same scale even a couple
266
years ago yeah only very few developers could produce the volume of code
267
that now every developer can produce
268
and they had to at least kind of understand it to do
269
that yeah there's a couple things going on here one of them is maybe this gets better as models
270
or harnesses get better but it does sound like understeered ai
271
if you just tell it sure man keep going for long enough without giving it feedback
272
or direction it does kind of go off the rails with current.
273
current tech
274
so that's that's i don't know operator skill laziness error like
275
they could get better at this bit the other bit is etiquette
276
or conventions or expectations or culture
277
and there's a there's a handful of good posts people have written some things
278
that i decided i agree with after reading their posts about
279
this one of them is called don't be a meat proxy oh yeah
280
and it describes exactly the situation where someone submits some code
281
with ai you review it you say these things need to
282
change they take your feedback type it into their agent
283
and say do this
284
and then you are operating their ai agent through them as a as an expensive fleshy proxy and
285
so they're not doing anything like you would it would be
286
faster for you to just talk to the agent right in the first place yeah don't do
287
that you're just a very expensive tube
288
yes yes a very expensive tube there's also the part of why this feels
289
so bad is because the reality used to be you couldn't really write something without understanding it right
290
so these artifacts carry this now not true connotation
291
that someone thought about this right and
292
so you're you're not the first person to try to uncover
293
some meaning out of it someone else did maybe it's weird
294
and wrong and bad but they at least exercise their brain enough
295
and now that's not true you are probably the first person
296
who has thought anything about the actual content of these documents
297
or code changes and that sucks it feels so bad it feels bad
298
but it does come with one good upside what's that
299
which is that you can you can critique the heck out of it
300
and nobody's ego is going to be bruised
301
because they didn't write it they didn't think about it yeah i mean
302
that is true it is easier for someone to say this
303
stupid robot did something instead of this stupid person did something
304
but it doesn't scale very well like the the meta feedback
305
has to be you should probably have read this first like don't don't make me read this
306
and understand it for the first time yeah exactly
307
because the feedback isn't actually about the code it is right
308
but at a surface level it is
309
but the deeper feedback is you sent you sent me crap
310
and you didn't even put in the work to review it
311
and now i'm mad
312
so now i guess we are back to the ego i
313
guess we're never going to get away yeah like you you
314
just produced a ton of work for me you it's uh
315
oh there's some concept here is it uh there's some like
316
ddos thing where you send out these little tiny packets
317
and they just expand what is this thing i can't remember what it's called i don't know it's it's
318
that thing the thing i can't remember oh yeah i know
319
what you're talking about just kidding i actually you just take this really small input
320
and then it it explodes and saturates usage somewhere Well, one specific example of that is a fork bomb.
321
Sure.
322
Yeah, you can think of it like a fork bomb.
323
Yeah, they fork bombed you.
324
They sent these little snippets of curly braces, and then that exploded into these tons and tons of changes that you have to review and understand.
325
Right.
326
And the effort disparity feels really bad.
327
It feels bad to be on the other end of that.
328
Yeah, it really does.
329
And I just don't understand why do people think it's acceptable to send code for review
330
that they themselves just have no idea if it's good or not?
331
I mean, they're literally asking you to do something that they themselves could not be troubled to do.
332
There's people that do this, right?
333
There are, yeah.
334
I mean, there are places that say we don't review code by hand anymore.
335
Humans don't review code anymore.
336
Yeah, but that's different.
337
Like, if you as a team agree we're not reviewing code as a team, fine.
338
But if you haven't made that agreement and you say, we do manual code review, what do you think is going to happen
339
when you send your code over that you haven't even read over to a reviewer?
340
Like, by definition, you're creating 10 times more work for them than you exerted to create it.
341
Yeah.
342
So, like, this just doesn't balance out.
343
Like, the scales are all messed up now.
344
I've also created this problem at my own company by getting non -engineers hooked up to coding agents.
345
And then they just, their job is not to understand code they are enabled by ai
346
but they i don't know we just don't expect them to be able to read it
347
and understand it but it is the same instance of this problem it's just culturally more acceptable
348
because they're not an engineer so we we say yeah we'll figure it out
349
but you haven't messed up by not reading it
350
and deeply understanding it right
351
but we don't expect you to it still is a bunch of effort right
352
but in this case you agreed to that effort yeah i feel like the social contract that's being broken here, is I never agreed to be the,
353
okay, I think I'm starting to finally form a concrete image of why this is so wrong.
354
So let's carve out exceptions for the case that you described, where you gave permission to non -engineers to produce code that the engineers said,
355
we'll review this and make sure it's good before going to production.
356
Okay, great.
357
Fine.
358
Contract in place.
359
everyone's happy i mean you're not happy happy
360
because your workload just went crazy
361
but at least you're not at least your expectations are being met
362
and then you have let's also let's also carve out an exception for the teams
363
that have agreed we're going to not human review any of
364
our code we have other means for quality checking the code before it goes to production
365
but we're not going to just human review okay carve
366
that out as well let's focus on this scenario where you say we are keeping our human review step
367
but there are team members who are not even reading the code the first time
368
so why is this a violation of a social contract
369
and why why does it feel so bad to me i think i finally understand
370
and that is that as the developer who's sending the code over for review
371
that they haven't even read you are outsourcing the pre -review
372
job to me review like let's break the word review down re means again
373
and view means to look at
374
and review by definition is someone has already looked at this
375
and i'm the second pair of eyes that is going to confirm
376
that it's good i'm confirming someone else's work
377
but this isn't code review this is actually code view like
378
no one it's the first time it's being looked at it's the very first time it's being looked at by human eyes.
379
And so the name is wrong.
380
The expectations are wrong.
381
And so it's, and also the reviewer is the wrong person to fix the problems.
382
I mean, think about how much work it is now when you say, look, the first time human contact is ever going to encounter this code is at the quote review stage.
383
Now the reviewer has to write all these comments down.
384
And where are they going to go?
385
Well, they're going to go straight into the LLM.
386
Well, then darn it, you should have had the engineer do that in the first place, because that's a much faster feedback loop.
387
So you're just operating in this totally inefficient way, and the roles, the responsibilities of the two players here are just completely misallocated.
388
It does make me feel like there's some steering or, yeah, you can, there's the cultural solution of saying,
389
hey, you still have to read all of your code before you submit it for review.
390
Even if the AI wrote it, you have to have looked at every single line.
391
And that's something you could bring up to the team and say, do we want to do this?
392
Yes, yes oh who said that weird sounds like there's a consensus
393
there's also some work you can do to kind of steer the agents
394
and how exactly this works best is changes all the time with new models
395
and harnesses and some combination of linting and
396
and guidance injected of here's how we do stuff this can help
397
but there's always things
398
that don't quite fit still it can it can reduce the amount of just totally off the rails wildly
399
incorrect things that happen that's probably not a thing you want to just go off
400
and do by yourself and the other part
401
that feels bad this did happen before lms i've seen engineers
402
that are sort of like the the defenders of quality right and they view themselves maybe correctly maybe not as
403
responsible for holding out the barbarian unwashed masses
404
that just want to ruin their nice code base yes they're going to do it the crappiest fastest way
405
and they have to hold the line right
406
and it is different in some ways because of the the robots
407
but that's usually a stressful spot to be in to feel like i have to like put up barriers
408
or the code will dissolve into into slop
409
and the business isn't going to like that
410
because i'm i'm making stuff go slower
411
And it just feels bad to be the person that tells people no all the time.
412
But you have the same problem at higher scale.
413
And one of the ways to get rid of that problem is to write stuff down and agree on it.
414
So to the extent that these are specific things, you can say, we don't do it this way.
415
It's worth doing.
416
But also certainly worth doing what you said, Dave, of the Benno level.
417
You have to have read this first.
418
If you don't, we'll close the PR.
419
We'll just say no and close it.
420
Maybe just put one checkbox on your PR template that says, I have read this code.
421
Yeah.
422
And the agent's going to check it.
423
Yeah, I read it.
424
I had to predict all these tokens.
425
Yeah.
426
The agent's like, yep.
427
Maybe we should have two checkboxes on every PR.
428
There's a code view checkbox and a code review checkbox.
429
I think we need three.
430
And then it's like the TCP handshake of like SYNAC.
431
Is that what it is? it's code view code re
432
and code review yeah those are the three -way handshake yes yep that's what we need
433
but the agent did the first handshake
434
or the first it just did step yeah the human just
435
did udp just said hucking this out there good luck everyone
436
else yep that's right that's not very polite yeah udp is not very polite well Have we answered it?
437
Yeah, I mean, I think at the end of the day, that's what this comes down to.
438
This is a process problem at the team level
439
that needs to be agreed upon that we are or are not going to do this.
440
And I think successful teams right now are doing a bit of a hybrid where they understand the scope of impact, the blast radius, the importance of different code.
441
And they know that some code needs human scrutiny twice, you know, and you used to get one by virtue of writing it now you get zero automatically
442
and so two human scrutinies is enough now having said that
443
you know maybe maybe we need to rethink this jameson
444
and say look
445
if you do insist on human reviewed code maybe the developer
446
who prompted the ai to write the code is actually the
447
reviewer like why are we inserting now a third essentially a third view
448
because the ai wrote it you know
449
and there's there's some amount of correctness to that and then the human developer reviewed the AI code, why can't it go to production now?
450
Why does it need another human to look at it?
451
You've already had three agents now at this point.
452
You've got the AI agent, the human agent, and then the human reviewer agent.
453
And I say agent kind of like in air quotes because humans are the original agent.
454
I don't know.
455
Maybe that's what needs to happen here.
456
But nevertheless, teams are figuring this out in real time.
457
And what we figured out on my team is that some code, yes, indeed, does need very careful scrutiny.
458
It needs to be babysat as it goes to production.
459
There's all kinds of alarms and monitoring you need to make sure it's not going to break something.
460
But some code, it's like, I just don't care that much if it's totally correct.
461
I also don't care if it's totally well architected.
462
And you have to decide which code that is for your team.
463
And then you really ought to agree on it.
464
Because at the end of the day,
465
this conflict is happening because of mismatched expectations on the team and an unclear process that has not been,
466
at least in the mind of you, has not been honored.
467
Or maybe a better way to say
468
that is two different team members actually have different ideas of what the process should be.
469
Yeah.
470
I have seen teams say, we focus on reducing risk by prompting in QA that runs automatically
471
and i don't know a whole bunch of technical things
472
and making it really fast to fix something if it does break
473
and then they can they they kind of yolo more stuff yeah
474
and then others that still say no like this this requires quality we can go faster
475
but we still have to understand what is being shipped yeah there's kind of a spectrum
476
but i agree
477
that you you probably need to pick one on your team
478
explicitly instead of have this fuzziness yeah i think that's right You're not alone, certainly.
479
I think you and the person you're talking about.
480
There are many instances of all of you, and some of them are the same people, actually.
481
I'm probably both these people.
482
I work sometimes.
483
At different times, yeah.
484
Yeah, but it's times they are changing.
485
Interesting times.
486
I don't think you are.
487
I think you are sensitive to what other people have to do in response to your actions, and I'm a little bit like that, too.
488
In fact, the other night - I did just think about our two last releases.
489
You broke both of them?
490
may have had a little whoopsie doodles sneak out in each
491
of them well i mean i've i left an agent running
492
overnight on a performance problem you know i wanted it to
493
improve the some some performance in a part of our application
494
and and it came back i woke up in the morning
495
and it had made a bunch of changes and i looked them over
496
and i'm like i really don't know
497
if these are good the numbers look good sure man whatever
498
but i i really would feel bad dumping this in my
499
team's lap you know they got a full workload already oh
500
you went the other direction yeah that's funny no i nicer
501
than i am yeah i just i mean it was it's a critical part of our system too
502
so i really want a human to look it over
503
but i really hate being the the boss who's like hey everyone drop what you're doing
504
because i had an idea yeah something crazy stand back everybody uh remember all those priorities we agonized over
505
and planned the whole quarter to lay out well we're throwing it all out the window
506
because i was in the shower this morning
507
and i thought wouldn't it be cool if
508
and now you get to review 2 000 lines of code
509
and then you just get to lecture everybody about how important
510
it is to be high agency you just do stuff the world belongs to people who do things that's right
511
and they're like okay i i do not review your pull request then
512
i use my agency to do the other stuff instead right
513
i think high agency means do whatever dave's intrusive thoughts say that we should do yeah
514
yeah oh oh anyway shower thoughts are they've never been more powerful
515
i know they can get legs a lot fast it used to be i'd have a shower thought
516
and i'm like that's gonna take me 40 hours
517
and now i'm like i'm running to my computer right now and typing that in yeah yeah
518
good luck everybody else yeah standing in the ashes of my
519
shower thoughts i may have i may have just shipped like
520
10 000 lines of terraform code last week did you really
521
some poor person on my team had to kind of sort
522
of review oh had to pretend to review yes they're like wow
523
that review only took you three minutes we had a gentleman's agreement where i said please review this wink
524
and he said great i will wink
525
and then we just didn't talk much after
526
that until the approvals came in he's like wouldn't you know
527
it 10 000 lines of terraform not a single problem found impressive uh yes yes you're like i'm
528
so good we are that good jameson's
529
so good at terraform it's those thousands of hours you've spent agonizing over terraform files finally paid off.
530
Yep.
531
Perfect Terraform first try.
532
Well, you said it.
533
We're perfect.
534
The question is answered.
535
What can people do if they want more of this perfection?
536
Yeah.
537
All you got to do is go leave a review for us, but not just make sure it's a good review because we're very sensitive and insecure.
538
Yes.
539
No, but actually, you know, I haven't checked on our reviews lately.
540
We should go read those and read some of them on the air
541
or something might be fun maybe oh maybe people can start
542
changing their apple podcast names like people change their patreon names
543
and we'll read the craziest ones oh man that'd be
544
so funny doesn't
545
that have to change your name in like yeah like messages
546
on your bank like your apple pay everything yeah yeah i
547
think i mean how committed are you for true believers
548
if you're really a soft skillet uh you will change your entire identity as recorded by your local government.
549
Oh, you're saying like your legal name.
550
Yeah, why not?
551
I don't know.
552
It feels like if you're going to go, you ought to go.
553
Yeah, be high agency.
554
Yeah.
555
And do the thing we told you to do.
556
Right.
557
Use Jameson's high agency.
558
Yeah.
559
Make me high agency by you doing the thing I want you to do.
560
Said every dictator ever.
561
All right.
562
So anyway, if you want to get your own question answered, go to softskills .audio
563
and click the ask a question button we have carefully prepared a form for you to fill out
564
and i gotta say all the code you see on soft skills audio was written by jameson's
565
and my fingertips oh actually we did have a consultant help at one point
566
so but that's pre -ai it's all artisanal handcrafted yeah this is handcrafted website it's very beautiful
567
and you can fill out our form where you can ask our question thank you
568
so much to those of you that do
569
that every week we read all your questions
570
and we love them i wish we could answer all of them in real time
571
but we will answer them on a delay basically what happens is we fall behind every week a little bit, but we'll make it up in volume through time dilation.
572
Right at the end, yeah.
573
Yeah, right at the very last moment.
574
Yeah, some kind of multiverse cloning thing will scale up.
575
That's right.
576
You know that scene in, what's the Marvel movie, Infinity War?
577
How Doctor Strange goes and explores like a million possibilities.
578
Oh, yeah.
579
And comes back and he says, only one.
580
That's us, but we're going to answer all your questions in those multiverses, universes.
581
Yes, I can't wait.
582
All right, thank you for listening.
583
We'll catch you next week.

Download App

Everything you need to speak fluently

AI PronunciationScore every sentence
IPA PracticeMaster every sound
VocabularyBuild your word bank
Vocab GameLearn while playing

Same Channel

FOR YOU

Recommend Video

Vocabulary and speaking notes for this lesson

This video has 583 sentences and 6272 words to shadow. The speech runs for 33:21. The speaker talks fast, about 188 words per minute, so expect linked and reduced sounds. 84% of the words are among the 3,000 most common in English; the rest is worth studying before you start.

Key vocabulary in this video

15 words from the video worth learning, with pronunciation and meaning:

WordPronunciationMeaning
feedback noun/ˈfiːdˌbæk/Critical assessment of a process or activity or of their results.
consultant noun/kənˈsʌl.tənt/A person or party that is consulted.
angel noun/ˈeɪ̯n.d͡ʒəl/An incorporeal and holy or semidivine messenger from a deity or other divine entity, traditionally depicted as a youthful, winged figure in flowing robes.
insecure adjective/ɪnsəˈkjʊɹ/Not secure.
developer noun/dɪˈvɛləpɚ/A person or entity engaged in the creation or improvement of certain classes of products.
gather verb/ˈɡæðɚ/To collect normally separate things.
expectation noun/ˌɛk.spɛkˈteɪ.ʃən/The act or state of expecting or looking forward to an event as about to happen.
legend noun/ˈlɛd͡ʒ.ənd/The life story of a saint (such stories are often embellished, but any kind is called a legend).
reveal verb/ɹɪˈviːl/To uncover; to show and display that which was hidden.
engagement noun/ɪnˈɡeɪd͡ʒ.mənt/An appointment, especially to speak or perform.
podcast noun/ˈpɑdˌkæst/A program, especially an audio program, produced on a regular basis, delivered over the Internet in a compressed digital format and designed for playback on…
cigarette noun/ˈsɪɡəɹɛt/A small cigar consisting of tobacco or another substance, wrapped up in a thin roll with paper, intended for smoking.
fork noun/fɔɹk/A utensil with spikes used to put solid food into the mouth, or to hold food down while cutting, or for serving food.
audio adjective/ˈɔ.di.oʊ/Focused on audible sound, as opposed to sight.
sensitive adjective/ˈsɛn.sɪt.ɪv/Having the faculty of sensation; pertaining to the senses.

Phrasal verbs you will hear

WordPronunciationMeaning
come back verbTo return to a place.
fill out verb/fɪl ˈaʊt/To complete a form or questionnaire with requested information.
go off verbTo explode.
mess up verbTo make a mess of; to untidy, disorder, soil, or muss.
bring up to verbTo raise (something) so that it reaches a particular standard or minimum requirement.
come across verbTo change sides; to cross over to work for the opposition.
come down to verbTo reach by moving down or reducing.
end up verbTo bring to a conclusion.

Pronunciation to watch

The speaker uses 183 contractions and reduced forms, such as don't, I'm, you're. Say them the short way, as you hear them.

  • The “th” sounds: gather /ˈɡæðɚ/, nevertheless /ˌnɛvɚðəˈlɛs/, therapist /ˈθɛɹəpɪst/
  • The “sh” and “zh” sounds: expectation /ˌɛk.spɛkˈteɪ.ʃən/, exception /əkˈsɛpʃən/, shout /ʃaʊt/, essentially /ɪˈsɛnʃəli/, interpretation /ɪnˌtɜːpɹəˈteɪʃən/
  • Long words — get the stress right: developer /dɪˈvɛləpɚ/, expectation /ˌɛk.spɛkˈteɪ.ʃən/, acceptable /əkˈsɛp.tə.bəl/, necessarily /ˌnɛs.əˈsɛɹ.ə.li/, complicated /ˈkɑm.plɪˌkeɪ.tɪd/

How to practise with this video

  1. Listen to the whole video once without speaking and note the words you do not know.
  2. Start at 0.75× speed, shadow it sentence by sentence, then go back to normal speed once it feels easy.
  3. Record yourself and compare with the original, paying attention to words like feedback, consultant, angel.

Grammar in this video

The structures the speaker uses most, with the exact words from the video:

StructureIn the video
Conditional sentences if + clause, will/would + verb — a condition and its resultif they don't, I'd be · If you don't, we'll close
“Used to” used to + verb — a past habit or state that is no longer trueused to be · used to get
Present perfect have/has + past participle — a past action that still matters nowhas relapsed · I've also noticed · hasn't updated
Passive voice be + past participle — the focus is on what happens, not who does itwas built · are involved · We've been undermined

What is the Shadowing Technique?

Shadowing is a science-backed language learning technique originally developed for professional interpreter training and popularized by polyglot Dr. Alexander Arguelles. The method is simple but powerful: you listen to native English audio and immediately repeat it out loud — like a shadow following the speaker with just a 1–2 second delay. Unlike passive listening or grammar drills, shadowing forces your brain and mouth muscles to simultaneously process and reproduce real speech patterns. Research shows it significantly improves pronunciation accuracy, intonation, rhythm, connected speech, listening comprehension, and speaking fluency — making it one of the most effective methods for IELTS Speaking preparation and real-world English communication.

Shadowing technique: read the full step-by-step guide →