In this episode, Kenny and Andrea discuss how to move from a blocking, "ivory tower" or hands-on architect role to a more facilitating one. They explore the importance of transparency as a first step in improving an organisation's approach to software architecture and design. The goal is to shift decision-making to the people who have the most information—the ones actually writing the software.
Benefits of ADRs:
Find a partner and begin writing ADRs for tough decisions in your project. Do it out in the open so others can see the benefits of the practice. This small, disciplined start can encourage others to follow suit, leading to broader adoption across the organisation.
Kenny (Baas) Schwegler: Welcome
back to another, well, whatever
2
:we call it, we don't know yet.
3
:Right?
4
:Andrea?
5
:Andrea Magnorsky: Yeah.
6
:Yep.
7
:Kenny (Baas) Schwegler: on facilitating
software architecture and design,
8
:and we have a new subtitle.
9
:Andrea Magnorsky: Yeah.
10
:And that subtitle is The
Realities of Systems Change.
11
:Kenny (Baas) Schwegler: Yeah.
12
:And unfortunately Andrew isn't
here for the second episode.
13
:He's still on holiday.
14
:Hopefully enjoying himself, but he'll
be here next time he promised us.
15
:So last time we did a small
introductions on on, on why are we
16
:actually doing this podcast or video
recording, blog, post series, whatever.
17
:and we talked at the end, like today we're
gonna go dive into transparency, right?
18
:So the stories are all
about how to get towards.
19
:Ivory tower or hands-on architect
being blocking to how can I
20
:actually let teams start to do
architecture and not be blocking?
21
:Right.
22
:one part you last time
mentioned is transparency.
23
:So what, dive into that.
24
:What do you mean with that?
25
:Andrea Magnorsky: Yeah.
26
:And also I wanted to add to,
it's not just tower architecture.
27
:That can be what?
28
:Blocking on the flow.
29
:It's also, even if you have a more
hands-on architecture, it, both of
30
:them can be blocking in some way
because at the end of the way of
31
:today, you're kind of relying on.
32
:A small group of people being the
ones that make the decision officially
33
:right, even if sometimes they don't
have sufficient context or whatever.
34
:And sometimes they do, but
you know, generally this is
35
:where the blocking happens.
36
:So we talked about the
side effects of that.
37
:Let's not revisit, should we.
38
:Kenny (Baas) Schwegler: No, and
I think a good book to dive into
39
:more is turn ship around, right?
40
:We want to make the decision to where
the knowledge and information is.
41
:So yeah, instead of being the hands-on
architect that has the information or
42
:the, the, the, ivory tower that has
the information, you think you have
43
:the information, but actually the
people actually writing the software.
44
:You have the information, you
want to turn that around and
45
:facilitate them making the decision.
46
:So that's
47
:Andrea Magnorsky: Yeah,
48
:Kenny (Baas) Schwegler: turn
49
:Andrea Magnorsky: teach them.
50
:Kenny (Baas) Schwegler:
yeah, we're gonna add
51
:Andrea Magnorsky: Yep.
52
:Kenny (Baas) Schwegler: yeah, transparency
is a very important factor, right?
53
:Andrea Magnorsky: Yeah,
54
:Kenny (Baas) Schwegler: be becoming
that facilitating architecture,
55
:Andrea Magnorsky: absolutely.
56
:So transparency is a great place
to start no matter what your
57
:organization like which part of the
journey your organization is at.
58
:It's a good way to start improving things.
59
:Because decision making
is a process, right?
60
:So if you're in a team, imagine you're
in a team and you're like, Hmm, we
61
:need to make some difficult decision.
62
:Or imagine that you are a principal
or an architect, a hands-on and
63
:you need to make a decision.
64
:Or if you're a Ivory Tower architect.
65
:Still, you have this, oh, okay, we have
some sort of problem or some request.
66
:You need to make a decision, and now
you need to possibly find some options.
67
:You know, you need to first agree more or
less on the problem, which it could have
68
:that that's never not always privileged.
69
:Sometimes it's straightforward,
but sometimes it's not.
70
:But then you have the option making,
and then you have, you know, hopefully
71
:you get to choosing one of the
options and you make the decision.
72
:However, all that process can, depending
on decision, take quite a while.
73
:And the suggestion here, or the
story comes to trying to bring
74
:transparency to that process.
75
:And that means that whoever you are
in all of this, you can, if this is
76
:not happening in your org, you can
literally start writing it down.
77
:And the whole point of
that is to just add, add.
78
:It's like having a memory of
what happened, or at least
79
:the way you saw it, so,
80
:Kenny (Baas) Schwegler: everyone,
everyone has that experience, right?
81
:Going into a project as a software
engineer and you're like, why did
82
:they, why did they make this decision?
83
:I think everyone has that and you
actually want to know what happened.
84
:So you, you become sort of like
this, historian looking at why
85
:did they make this decision?
86
:And I think that's, that helps, right?
87
:That transparency to just look at,
okay, why did they actually make it?
88
:Andrea Magnorsky: Yep.
89
:Kenny (Baas) Schwegler: What
happened in a conversation that
90
:they switched from A to B maybe, or
91
:Andrea Magnorsky: Yep.
92
:Kenny (Baas) Schwegler: why haven't
they thought of option like this?
93
:Oh, they thought of it.
94
:written down now.
95
:So everyone has that problem, but nobody
seems to, not nobody, but it seems to be
96
:very hard to actually start doing that.
97
:Andrea Magnorsky: Sometimes you're lucky
and you have someone that was there
98
:and the problem is that if you do get
a story, that story might just not be
99
:complete enough, one easier way to start
doing this transparency thing for your
100
:future self is to write decision records.
101
:And so I heard about decision records
many, many years ago, and the first
102
:time I experienced them, it wasn't a
great experience, and that's because
103
:we were told that we had to do it.
104
:Right.
105
:So sometime at some point it's like, you
know, some reason, I can't remember what,
106
:what was the reason, you know, you need
to start doing this decision record thing.
107
:But that was it.
108
:It wasn't like
109
:Kenny (Baas) Schwegler: Or like
a bureaucratic process, you
110
:Andrea Magnorsky: I.
111
:Kenny (Baas) Schwegler: do this else.
112
:You cannot continue.
113
:Andrea Magnorsky: So, to be honest, I
can't remember if that was the ethos
114
:of or the origin of this request.
115
:And it might have been well intentioned,
but I think writing decision records is
116
:actually kind of not, it sounds easy.
117
:And when you read the history
of how it came about with the
118
:Nygard what's his first name?
119
:I forgot his first name, but the,
120
:Kenny (Baas) Schwegler: Yeah.
121
:Andrea Magnorsky: Michael Michael Nygard.
122
:So.
123
:He's said, Hey, really simple.
124
:Just write it down, you know type it.
125
:That'd be all good.
126
:But the reality is that now that's
he, he presented, or at least the
127
:way the model I had for decision
records is that you write them
128
:at the end, you decided on some.
129
:Database choice.
130
:Let's say for example at the end
when you finish, which actually
131
:probably took quite a little time,
you're like, we chose Postgres.
132
:Reason one, reason two, and you
put your banger in your rebook.
133
:But that's, I don't find that
to be the most useful way.
134
:What I find more useful is
to write a decision record.
135
:Process that you start from quite early
in the decision making process, and
136
:then you start adding all the options.
137
:And it doesn't need to be a long document,
it just needs to have the information,
138
:Kenny (Baas) Schwegler: Yeah.
139
:Andrea Magnorsky: the end.
140
:Kenny (Baas) Schwegler: I don't
necessarily think the, I do
141
:wrote ADRs, but they weren't
enforced on me and it quite worked.
142
:the problem I had with it
is that it's after the fact.
143
:It's like writing your unit
test after your code worked.
144
:Right.
145
:for me, what you're saying as
well, a, it's transparency.
146
:If we dive into the transparency part,
what I see usually happening in a
147
:lot of companies now is that nobody
knows who can make the decision and
148
:nobody knows what has been said.
149
:And I talk to person A and they say,
blah, blah, blah, blah, and then I go
150
:to C and they say, yeah, but that person
said that and that person said that.
151
:So just writing the a ADR itself for
yourself and just say, well, you know,
152
:I've written down what that person
said, and that's this and this and this.
153
:You, you make it transparent
what people said.
154
:Of course I check up with,
with the people, right.
155
:Hey, I've written it down.
156
:Is this correct?
157
:Sure.
158
:Can I share it?
159
:Sure.
160
:And then, right.
161
:You want to have that consent
that it's shareable, but then the
162
:next time and in a meeting, well
actually the person said this and
163
:Andrea Magnorsky: Hmm.
164
:Kenny (Baas) Schwegler: oh, what?
165
:But we had a whole different conversation.
166
:I see it differently, which can happen.
167
:Sure.
168
:Okay.
169
:Then we sit down together.
170
:Right.
171
:So,
172
:Andrea Magnorsky: absolutely.
173
:Kenny (Baas) Schwegler: yeah.
174
:So I think that's the bit
one about the transparency.
175
:Just doing that already changes
what you were saying a lot.
176
:Just start writing them.
177
:Once you start making the decision and,
178
:Andrea Magnorsky: so one thing that
I find really useful is to if you're
179
:doing it as, some sort of group, because
imagine that there is some group of
180
:people and no matter what, your role
is in that and you decide to start
181
:writing this, but nobody else does.
182
:Not very useful or some better
than nothing, but still,
183
:you know, could be better.
184
:What I find more useful is if you
decide as a team, and again, starts
185
:on like start at a team level, not
like certainly everyone writes ADR's
186
:without any help or coaching or anything
more likely to fail, but if you just
187
:start with one team and go right.
188
:Let's write these ADRs and let's
have a decision record review every
189
:week where we go through the decision
records that were written this week.
190
:And the first ADR can be should
we use ADRs and the options for
191
:it and everything, which is nice.
192
:'cause what you, what
you wanna do is kind.
193
:Start doing the practice, right?
194
:So we're talking here about the,
how you need to facilitate the
195
:practice of architecture and, and
I think this is a prime example
196
:of something that's simple, but
197
:if you really foster it and enable it
about like what is a good ADR The fact
198
:that you should include pros and cons the
fact that you should it's very useful to
199
:include the opinions of different peoples.
200
:And this is really good advice.
201
:Like if you want an excellent summary
on what is a good decision record
202
:process that is in Andrews book,
which I hope we can discuss next
203
:time as, as well, when they are here
204
:Kenny (Baas) Schwegler: because I
like the way he adds the advice to it.
205
:Right.
206
:But
207
:Andrea Magnorsky: yes.
208
:Kenny (Baas) Schwegler:
come up to a good point.
209
:Whenever I talk to people like
this, they're like, oh, how
210
:do I start, how would I start?
211
:Well start small.
212
:So you were saying alone is okay.
213
:But I still think two people
doing it, you need at least two.
214
:It's also something that I was
researching cognitive sciences, right?
215
:If there's a norm and you're off the
norm, you think different, all you need is
216
:another person that has the same thinking.
217
:And if you both do it,
then there might be change.
218
:Andrea Magnorsky: Yeah.
219
:Kenny (Baas) Schwegler: need to.
220
:And that's what I actually started doing.
221
:I did it with someone else and we
just started doing it out in the open.
222
:We also had the facilitating
architecture role, right?
223
:So I was not in the team.
224
:What happened was very interesting
is that they saw it started to work
225
:because we kept referencing and we
kept doing the discipline, and all
226
:of a sudden other people started.
227
:Doing it because they
say, well, I've had it.
228
:seemed to work.
229
:I'm gonna write one now.
230
:And that really, that was really helpful.
231
:And all of a sudden it was all out in
the open and people were talking about,
232
:well, have you seen what they wrote down?
233
:Especially the advice, right?
234
:People writing the advice,
start attacking people in it.
235
:Like you need to write
down your advice here.
236
:We just talked on the hallway, but
it needs to be in this document.
237
:I've had it with all these cheaty
chat and this gossiping or all this
238
:back and forth, just put it in that
document and it's just a shared
239
:mental model and just two starting it.
240
:Together, out in the open for
tough decisions and making the
241
:decision actually work and people
being happy that set it off, right?
242
:So if you are in either what you're
saying, right, if you're in a team and
243
:you're just two people starting it, or
you are an architect between the teams
244
:and you start it off, try to do it down
in the open, experiment with it safely.
245
:I think that goes with a lot
of architecture tools, right?
246
:It seems usually so big, but
you can actually make it small
247
:and start experimenting with.
248
:Andrea Magnorsky: I think one
thing totally all of that and
249
:the, the thing that really.
250
:Helped in this case that I'm thinking
about is not just doing it in the
251
:open and kind of coaching through,
like in, in stand up, be like, oh, we
252
:need to, oh, I'm, I need to do X or Y.
253
:It's like, oh, that sounds like
you need a decision breaker.
254
:And people initially
were like, okay, fine.
255
:But, then they started saying it, oh, it
looks like you need a decision record.
256
:And that's when they started really
seeing the value because I think one
257
:of the main things that it does is
when you're having the review meeting,
258
:the review meetings are super focused
because if people that are there haven't
259
:read the decision record, you're gonna,
you start it off and you go like.
260
:Have people, has everyone here read this?
261
:If people go like, yeah, it is
like, okay, let's just spend
262
:five minutes and read it quietly.
263
:So you read it quietly and, and
yeah, it would've been nicer if
264
:it was read before, but you know.
265
:We are where we are.
266
:So read the thing and then actually
have a discussion about it.
267
:The minute this conversation
or someone is like, oh, but
268
:what about this conversation?
269
:You can easily say, and
this has happened as well.
270
:It's like, great question.
271
:Sounds like you need a new
decision record or whatever.
272
:It's just not part of this.
273
:And also, during the decision
record, everyone can collaboratively
274
:edit this decision record.
275
:So someone is saying something about this,
it's like, oh, you know, Kenny said X,
276
:Y, Z, and it can be there and then, and
so it becomes this really rich thing.
277
:And the best thing about it all is
that you have less of the same meeting.
278
:So generally when decisions
are difficult, you.
279
:Wind up having the same meeting
like a bunch of times, or people
280
:kind of think it's not, no, it
doesn't happen, but it does.
281
:And some of that comes
to how we make decisions.
282
:It could be that this team is
used to the comfort of consensus.
283
:Some other teams are the comfort
of delegating the decision and
284
:somewhere in between and even kind
of surfacing us, like who should.
285
:Make this decision, but now
it becomes an explicit issue.
286
:It's like, for this decision,
maybe we're okay deciding, but
287
:this other one we're not sure.
288
:And then actually you're like,
oh, we don't know who should
289
:make like, and then you start
thinking, what are the side effects?
290
:What's the worst that can happen?
291
:And you start maybe going on,
like it's an explicit question.
292
:So it's also a decision.
293
:So it's transparency or rank.
294
:Kenny (Baas) Schwegler: Yeah.
295
:And I think for the next part because
I think this is very relevant.
296
:So one thing of today's.
297
:Thing to get with start writing,
at least with, find someone you
298
:can start doing writing ADRs with.
299
:Use them for the whole process.
300
:Not at the end.
301
:And I think that that's
a you shift, right?
302
:Start it once you have the problem,
name it after the problem, and
303
:then start adding it while you
are trying to make the decision.
304
:And in the end you can do sort
of like, and that's what you were
305
:saying, like a refuse session.
306
:Now we should talk about this
in the next session, right?
307
:So about.
308
:Who makes a decision?
309
:What's his review session?
310
:Because that can also sound like
a sort of blocking gate keeping
311
:thing, but it shouldn't Right?
312
:It, it should be an advice thing and
not like we should discuss this first.
313
:It should be about learning
that, moment in time.
314
:There's many ways of doing that.
315
:Let's keep that to
another session as well.
316
:And one honest thing, I was thinking
about what I said earlier you are
317
:saying, well, don't just write them.
318
:Once you make the decision, you
write them for the whole thing.
319
:In the next session, I want to
invite Diana Mont our friend, right?
320
:Writing is thinking.
321
:she talks a lot about that and I
think it's the same with writing
322
:unit tests and then make the code.
323
:I think it's the same for me and I
want to explore that with Diana next
324
:time in one of the next session.
325
:Right?
326
:For me, it's the same like if you use
the A DR as a writing, as thinking tool.
327
:Instead of, we're just capturing
the decision that is a whole change.
328
:So maybe we can pick
up on another session.
329
:Yeah.
330
:Andrea Magnorsky: That
sounds like a great idea.
331
:I think for different
people it's gonna be.
332
:Be different.
333
:Like for example, I find modeling
really good for thinking.
334
:I find that sometimes I will write test
at the start and other times I don't.
335
:And it depends when I
do that, for example.
336
:So I think it's good to, I think
we're quite aligned, but not the same.
337
:And I think this is what makes
this conversation better.
338
:And also I think that when we're
talking about these experiences about.
339
:This experience I just talked about
and the one that you talked about,
340
:what I'd love to do with this series
of videos is that to help people
341
:that are anywhere in their, like
they're somewhere in this journey.
342
:That maybe they can take something
that is useful now and something
343
:that, wouldn't it be nice if it
was in some new way in the future?
344
:So to, it's like you're somewhere.
345
:Hopefully you take something away
and you can practice it, and then you
346
:come back and tell us how did it go?
347
:You tried something.
348
:I love to hear about that.
349
:Kenny (Baas) Schwegler: Yes, thanks.
350
:So I hope people enjoyed this
second episode next time with
351
:we'll have Andrew hopefully here.
352
:And please join us at the virtual EDD,
but you can also join us in Discord.
353
:would love to hear your comments
and feedback see you next time.
354
:Bye-bye.