How to Break the Rules

Dan North

Recorded at GOTO 2017

good afternoon okay
so I'm doing three really dumb things
the first thing I'm doing is a Katherine
asked me to to be on a track she's
talking about like IT technology and
society and how society kind of relates
to technology and how we apply it and
all of that I thought that sounds like a
fascinating topic I have no idea what
I'm going to talk about so so I thought
I'll need to put a new talk together so
this is a new tool something I'm doing
the second dumb thing I'm doing is
following Linda rising so if you're in
Linda risings talk that was amazing and
the third dumb thing I'm doing is
standing between a roomful of Dutch
people and beer right so so I know I've
ended that couple of other dumb things
which I'll introduce as we go but let's
start there first a word from our
sponsor here we go there's an aptly last
question throughout session and then I
would of course run out of time at the
end so you answer any of them but at
least you get them with a wonderful
internal glow of having posted the
question okay I might tweet some answers
so this talk then um I was thinking
about particularly with technology in
society how we as society how we as
human beings adopt technology and
there's something that fascinated me
recently probably in the last year or so
um Chuck called Ellie Goldratt Eliyahu
Goldratt died a few years ago sadly
massive brain very very smart guy so he
was a physicist and he was also probably
one of the most significant management
thinkers of the 20th century and he
wrote a a seminal book in 1984 called
the goal and this it was one of a around
that time there a couple of books came
out there's this one and a book called
zen and the art of motorcycle
maintenance by an English teacher
philosopher called Robert Pirsig and
they were novels so like business books
philosophy books presented as novels
and they had real trouble getting them
published so so Goldratt was quite a
successful business consultant by this
point and he wrote this this novel and
it's about a guy called Alex who works
in a factory actually runs a factory and
the factory is going down the tubes and
a legit Alex his marriage is on the
rocks and it's all doom and gloom for
Alex and he meets up with an old school
teacher called Jonah and Jonah kind of
coaches Alex in terms of figuring out
why the organisation's not working
terribly well and it all ends up OK in
the end so don't worry ok this is not a
depressing story um so and it led it
kind of led the led the groundwork for a
whole load of you get a lot of these
sort of business novels now so the most
obvious one is a Phoenix project which
was written as a direct homage to the
gold that kind of brought forward 30
years to the devops world well there's a
whole load of these business books now
things like five dysfunctions of a team
where you start with like a fable and
then it unpacks a fable so the goal was
a really significant book and it
introduced this idea of Theory of
Constraints and theory of constraints is
about how is it's it's how you can apply
lean operations and lean thinking to
improvement in the organization I don't
want to talk about that today I met a
chap called Kevin bear who's one of the
co-authors of the Phoenix project a few
years ago very very smart very
personable human being and he said to me
if you like the goal because I'm you
know bouncing around and massively into
this whole Theory of Constraints thing
he said if you like the goal you're
going to really like beyond the goal I
was beyond the goal I mean so well it's
a I went when he told me I thought I
heard book but it's a series of lectures
there's eight lectures that Goldratt
gave 21 years later so he's written this
book you know soso Theory of Constraints
lean operations isn't eating the world
20 years later and he's a little bit
perplexed by that so if you look in fact
this version of the goal the latest
version of the goal has a wonderful
extended interview with him in the back
who talks a bit about that but so beyond
the goal is this series of lectures that
he gave and again he dies like six years
later this is towards the end of his
life and oh my word if you think the
goal is like everyone has
who's ready to go this is going to have
okay so about 10% of you in this room
everyone else your homework is to go and
read the goal it is wonderful it's
life-affirming it's brilliant once
you've read the goal in about two years
time once you've really internalized it
then you're probably ready for beyond
ago so the only goal is and he's an old
Jewish guy and so he speaks like an old
amazing at telling stories
because old Jewish guys are amazing at
telling stories and he laughs at his own
jokes because he does he's got a very
kind of life into humor and it was
little just cheeky kind of you know
indulging himself laughing it's joyful
to listen to and so what he does in
about the first five seconds of this of
this first lecture is he drops a
and you think I'm quite good at this
stuff and then you listen to him and
your brain explodes or my brain exploded
and I said I'm just going to make a
statement and he makes his statement he
says he says technology can bring
benefits if and only if it diminishes a
limitation he says all I'll say it again
this is oh but I'm not very good at
judgement voices B so I'll say it again
technology can bring benefit if and only
if it diminishes a limitation and then
he proceeded to spend the next like two
lectures two hours of this eight-hour
thing unpacking that sentence and what
happened with my brain melted and I
realized what he's doing is he's
articulating beautifully why we really
struggle to adopt technology and so what
I want to do today is try and unpack
some of that for you so what I've done
is he uses two examples which I want to
give you
I've been added another two so I
basically doubled the amount of content
I'm going to try and do that in about a
half the time which means I'm going to
speak quickly ok tell I'm English and
that's what I do so if you struggle with
me speaking English quickly do you put
your hand up I'll ignore you but at
least you'll know that you know you put
your hand up so technology can bring
benefits if and only if it diminishes a
limitation so let's take a look what
does technology what is technology so of
course I did what anyone does I jumped
on the googles that's a dictionary calm
that Sumerian words
and technology is a really big paragraph
about stuff okay it's quite hard to
describe um however both of these things
have a few keywords in common they both
talk about knowledge so technology is
about knowledge and most importantly
it's about the application of knowledge
so if I just study something that's
philosophy okay if I then apply that
knowledge to something that's technology
okay so the practical application of
knowledge accomplishing a task using
with life so you're actually applying it
to do stuff so okay here's what I
concluded from listening to Ellie
Goldratt for a couple of hours we're
really really bad at adopting new
technology okay
that's generally my conclude in fact
adopting new technology we love new
technology we've got this whole the
lovely adoption curve of you know early
adopters and pioneers jumping on these
things and then the early majority come
along a bit later the thing we're really
bad at is exploiting new technology okay
we're rubbish at exploiting new
technology what does that mean what I
want to do is give you a series of
examples of technology advances over the
last probably 50 years and unpack those
a bit and essentially what what Goldratt
says is this is he says when we have a
new technology there's four questions we
need to ask ourselves in order to figure
out how to exploit that technology and
we need to ask these questions in order
and we need to answer these questions in
order and if we don't do that we will
not get the benefit from the technology
so these four questions unpack see if
and only if ok so you're going to write
anything down this is the stuff to write
now the first thing is this what is the
power of the technology what does it do
and he says this is the easy one because
there's people selling it over listen
read the sales brochures watch the
advert so that's what it does
ok it makes your wipes a new bluey wipe
that'll hold up against the window or it
makes sure
car go faster or it makes the journey
smooth or it makes whatever it is it's a
phone but without any buttons because
what we need is a phone with fewer
buttons okay the power of so technology
what what benefit does it give us then
he says well this is the a fun Ania fit
what limitation does the technology
diminish if we can't identify what
limitation the technology diminishes all
the other bets are off so there needs to
be a specific thing that says right now
we have this limitation this thing
that's weak with us that's a problem for
us and we have to be able to demonstrate
that this technology diminishes that
limitation this is surprisingly
difficult and the reason this is
surprisingly difficult is because mostly
those limitations are hiding in plain
sight we've gotten used to them it's
like arts going to fish house the water
and fish is like water water yeah
because it's in it's around it all the
time and again as I said I want to give
you a few examples and unpack that a
little bit so what limitation does a
technology diminish now it gets
interesting because now we ask ourselves
what rules enabled us to manage that
limitation we've had that limitation
we've always had that limitation we need
rules in order to manage that limitation
so what are those rules because if we
don't dismantle those rules we don't get
any of the benefit of the technology
because we're still constraining
ourselves by rules that work right now
for the thing we used to have a few
people nodding already yeah it's sort of
trying to apply this what the problem
with any Goldratt is he says these
things and they hit you like a punch in
the stomach going to hit like a sack of
sand and you suddenly and it might be
confirmation bias but you suddenly see
it everywhere you know one of the things
that a lot of the work I do is going
into organizations and helping them see
things differently ok the amount of
constraints I run into that are exactly
this we have rules that enable us to
manage the limitation and we are wedded
to those rules and finally well we've
introduced this new technology and it's
going to give us some new capability ok
we're going to need some new rules we're
going to need some new rules that enable
us to
manage this new thing okay so and these
new rules aren't usually they're not of
the existing rules they're different
things so this means we have some work
to do so I can't just go today here's a
new technology his electricity now you
just go and electricity all the things
and you'll be fine
yeah the agricultural revolution is
farming right well so if you approach
and gathering ain't gonna work out so
well you need to think about things
differently and likewise if we go into
the Industrial Age which we did thinking
it was just like the pre-industrial age
everything was really difficult for ages
well we figured stuff out then we
industrialized and then we got very much
into your scientific knowledge
management from the early 20th century
yeah Frederick Winslow Taylor and Henry
Ganz and Henry Ford and all those very
guys with very big brains and then
post-industrial were carried on using
those models okay so this is how you
adopt a new technology let's take a look
at some examples who's heard of MRP this
is an age test who's heard of MRP okay
two hands was two more than I was
okay this is probably one of the first
applications of computer technology in
business or certainly one of the first
widespread applications of computer
software packages in business MRP these
materials requirements planning is what
it stands for
if you're in a factory if you're in any
manufacturing context and you build
stuff for customers then you have an
order book and you have customers and
you need to build those orders you need
to fulfill those orders in order to
fulfill those orders you need material
now if you over order material you have
a sunk cost problem okay you have an
inventory problem if you under order you
have a resource starvation problem so
getting the right amount of materials is
absolutely critical to running a safe
cash flow well running manufacturing
business and this was a huge huge while
still is occasional this was a huge deal
in the mostly manufacturing late
twentieth century so fifties and sixties
and seventies in the Western world
so what we have is in a typical factory
you might have 20% of the workforce work
very very smart highly paid number
crunchers and they would be off in
losses somewhere doing the materials
resource plan there's material
requirements and the material
to calculate the material requirements
for the next month okay and so what we
did is we did it every month and it
would take a week to to calculate
painstakingly because it was error-prone
it was manual and all those things that
the material requirements for the
following month if you got it wrong you
either had way more spend than you
needed or you suddenly were unable to
fulfill orders both of those are really
really dangerous especially in a low
margin manufacturing scenario so then a
bunch of clever folks said we can make a
computer do this and so they did and so
they invented this in the MRP software
that was the first wave of like package
software and DuPont was one of the first
the very early adopters of this and
DuPont went this is awesome right we can
now literally run our MRP calculation
overnight and because they could run
one of the largest manufacturing firms
in the world certainly in the u.s. over
they were huge still are pretty
significant and so then what happened
was loads of other companies went well
look we need to compete with the likes
of DuPont what's their secret and they
said Oh MRP you should get some MRP
software know what brilliant and they
all adopted is MRP software and they're
like nothing's happening we put the
pixie dust on it nothing's happening why
is nothing happening and so you can look
at this scenario this is before my time
it is an le Goldratt story I have my own
stories but he said he lived through
this and he was consulting in two firms
while this was going on and he didn't
recognize this it's only in retrospect
this is like 20 odd years after this has
all happened he's going I get it I see
it now he says let's unpack this MRP
scenario using these four questions so
the first question then what is the
power that MRP systems provide what is
the power of that technology what does
it allow me to do
right we've really really complicated
calculations I can do them overnight
which is amazing right and carry out
complex MRP calculation and they're done
by the following morning okay so what
limitation then what limitation does
this diminish yeah so the limitation it
diminishes is right now it takes an
entire week for a skilled team to
calculate MRP for just for the next
yeah it's time-consuming it's expensive
its error-prone I probably have teams
doubling up on calculations because
there's lots and lots and lots of
paperwork involved so then what are the
existing rules that made it possible for
me to do materials resource requirements
planning in a large factory what rules
did I have that said this is what we're
going to do are the rules here we can
only plan monthly that's one of the
rules right planning more than monthly
right is risky okay but planning less
than month is too expensive I can't do
it yeah
so monthly is about the right cadence if
I'm selling a week a month that's
probably about right anyone who's using
scrum right spending like one of the
people I was working with on recently
and I said oh you know how you finding
with agile stuff is all using scrum he
said instead of doing my head in
I said why I said all the meetings I'm
like but it's it's agile you don't have
meetings it goes I don't call them
retrospectives and sprint planning and
backlog grooming and if the word meeting
isn't in there but all meetings and you
know a significant chunk of my week is
in meetings I'm like yeah you nailed it
you're absolutely right
so that means that you can't have
shorter Sprint's than that because it's
just too expensive they figured out the
same thing okay takes a week a month to
do the planning and we need to buy a big
enough batches I need to order big
these were the rules that I had in place
so now what new rules are going to
enable us to exploit this technology
enables to benefit from this technology
and and so well the first thing is we
need to replan frequently so Dupont
realize that if I could make these
calculations overnight I could basically
order overnight which is pretty cool
I could certain the order much more
and also I can order in shorter
timeframes which means I can order sure
smaller batches which means say there's
something wrong with the batch there's
less goods to send back so I start
creating this win-win-win a smaller
batch size this is an accident a happy
accident but then that means a bunch of
things in order to exploit this
technology we need to change the
relationship with our suppliers right
and with our customers in terms of our
suppliers I need to say can you deliver
smaller batches more frequently right
now you have a massive series of trucks
that turn up every you know 30th of the
month or whatever you know - my goods
for the next month what I actually want
now is can you send like very small vans
every couple of days and for a lot of
their supply chain that was completely
mind blowing for them they had to retool
and rethink a lot of things and a lot
forget is only DuPont they're doing this
and everyone else isn't then basically
the markets pretty flat there's no ones
no one's at risk because they're not
keeping up it's only as you cross that
chasm and now some of the earlier the
early majority late majority start
adopting these systems that the laggards
are now exposed as not being able to
respond and I can also change a
relationship with my customers
hey customer you can change your mind
you can order in smaller amount so you
want less value at risk you've got less
sunk cost okay as you realize whether
your customers are but your customers
are buying product or not you can change
the size the configuration all of that
kind of stuff yeah in the late eighties
early 90s
Dell computing created a multi-billion
dollar business model around this so up
until Dell you basically when you bought
a computer it was it was the Henry Ford
you know any color you like as long as
it's black yeah HP Compaq all the big
vendors IBM are saying this is what a
computer looks like and Dell were like
how much memory do you want how large of
a hard drive how big of a screen all the
other configurations and you just had
sliders and used Moodle the sliders
Newman chip and it would arrive and
Italy arrives before your order would
arrive anyway from the other guys so
they had this whole supply chain thing
down and they became you know Michael
Dell became a billionaire because he
could work in smaller batches right so
we need to change a relationship with
our supplies and our customers
once we do that we can outmaneuver our
competitors this is pretty cool but
until we do that we are basically
constrained by exactly the same rules we
were using before and this is what was
wrong footing all of DuPont competitors
as they were using these systems and
they were still doing their MRP
calculations every month every month
they would run them overnight and then
the following month they'd run them
overnight again so they were getting
this so Bret is much cheaper we've got a
huge cost saving we don't need these
rooms full of very expensive calculating
human beings anymore but we're getting
none of the business benefit we've
managed to save a few bucks on salary
brilliant our du Pont's busy changing
how manufacturing works yeah because
they were still clinging to their old
rules so the first kind of takeaway I
got from this is this all those rules
that we use for coping they become
policy and once those rules are policy
it's really hard to unpick those rules
okay because different people own those
rules and different people kind of
collect around those rules so the policy
is we plan monthly there's a monthly
meeting that we have that's policy big
enough batches is policy so you got run
who works in the warehouse and he's used
to the trucks coming in and you know
what he's quite slow-moving and quite
ponderous and very thorough and always
check the paperwork
you've got vans coming every couple of
days he's really unhappy about that okay
so we need to bring wrong with us as
well and if idea that we need to change
the relationship with our suppliers and
their customers if we don't think to do
that again we get none of the benefit oh
okay so let's take another example
ERP so who who's heard of ERP all the
variegate so that's a generational thing
right that's seventies and eighties and
nineties okay so a lot of people in this
room grew up with very big very slow
moving ERP system ERP was another
seismic shift in enterprise certainly
ERP was the idea that organizations have
data and that
is usually in silos and because that
data is in silos it's expensive and slow
to obtain that data and so then we kind
of shore up our world and say well we
can't really move quickly what we'll do
is we'll make the best decisions we can
given what we know so the sales silo
makes a sales policy decisions and the
manufacturing of engineering silo makes
engineering decisions and the production
silo and the sale so all these different
silos are all busy doing their thing ok
ERP comes along and says in much the
same way as MRP install this systems you
can now have access to real-time data
about stocks about sales about
utilization about people about movement
of staff everything all your business
metrics that allow you to make good
commercial decisions good operational
decisions are at your fingertips you
just need to pave you know 15 million
euros and get out a data center and have
you know 28 ASAP consultants in your
building for the next four years and the
rest is grazie it's easy yeah Wow I want
some of that but start 12 million euros
I stand to save a hundred million euros
this is chump change your go and then of
course story after story of
organizations who took on it's huge
systems and huge transformations and
nothing really changed and they're like
oh ok we've sunk 12 million in and
current calculations always saved 489
euros and supply chain yeah great and
how much did you spend on that yes yes
basically the farm right so ok we can
unpack this in terms of how do those
four questions land so the first
question what's the power of ERP I can
now collect I can process I can present
information across the enterprise all
three of those activities are really
hard collecting data in any kind of
real-time is really really hard in a big
organization manually ok and also people
make stuff up right because they tell
you what you want to hear rather than
what you've watched you should really
hit and across the enterprise is now
we're going across silos that was a huge
deal so ok what limitation then does
ish right what's that what's the sales
pitch what's the benefits case well so
the real limitation is ignorance it's
ignorance of what other groups in the
enterprise are doing and so I need to
make local decisions so what kinds of
things did we do before ERP this is a
really tough question because the answer
is all the things all the things you do
in an organization that are around
process governance oversight program
management you name anything you do
that's about information which in
knowledge work is most of the things
and in fact enterprise is most of the
things are about having to make
decisions so the most obvious one means
we focus on utilization okay Donald
Reinertsen says wonderfully so the bad
models better than no model so the model
we're going to go with is utilization
over all these expensive people I hope
they're expensive right all these
expensive people and and I need to well
- I need to sweat my assets right I need
to keep them all busy so what I want it
for all of you to do a timesheet and I
want you to tell me exactly how you're
spending every single minute of your day
you know and then one of the things that
people's time it gets allocated to
filling in timesheets yeah but so I need
to know where my spend is going okay and
so I want to check utilization and then
we invent a thing called cost accounting
this it's not an enormously intuitive
thing yeah cost accounting says if I can
maximize utilization for my assets then
I'm getting the most value I possibly
can from my assets so I think of the
world is static and largely hierarchical
and if I have someone who's a head of
sales and he's responsible for my sales
activities and he says okay well you
need to maximize your sales assets and
he has a bunch of sales managers and
they need to maximize their teams right
away down yes to the plebs on the ground
and then she she's my production manager
my production people so that's what cost
accounting does it says I need to make
decisions I don't know what sales is up
to on production I don't know what sales
is up to I don't know what this batch is
up to I don't know what marketing is up
to because I doesn't have access to that
information I can ask my peers but that
informations in a weeks or months out of
date by the time it filters up the chain
of command so I need to make local
decisions and I do that using a whole
bunch of modern models called cost
accounting and so that means now that we
start getting this really perverse
organizational thing where anything that
isn't sales is a cost center okay
so IT is a cost center engineering of
the cost center production is a cost
center anything that doesn't involve
someone actually transacting cash coming
into the business is seen as a cost
center and so because I don't understand
how value works in my organization the
only lever I've got is cost so I Drive
that down and what happens when you
drive that down all the bad things so
now what can I do
what new rules do I need with the new
rules I need are about a thing called
throughput accounting so again only
Goldratt describes this in the goal
there are only three numbers you need to
run a business there are only three
accounting numbers you need to run a
business the first is called inventory
inventory is how much money have I spent
on things that I intend to sell so all
the bits I've bought that I'm then going
to do stuff to and sell that's my
inventory okay throughput is sales write
throughput is I've now built all these
things and I'm shipping all these things
and people are paying me money so
throughput is all your revenue and
whatever else and the other number
operating expense so operating expense
is how much does it cost me to turn my
inventory into Troopa it sounds
incredibly simple because it is right
the problem with it is you need to be
able to see the whole system in order to
be able to answer those questions if the
system is for people sitting in a garage
is easy to calculate inventory operating
expense and Trouper if the organization
is 3,000 people it is well nigh
impossible so you've already do is you
fall back on a bad model you fall back
on cost accounting and we want to
measure flow of value so that means how
work is happening and how it helped
happy providing value we need to do that
holistically across the whole
organization that's the first thing we
need to do weirdly and this is
countercultural okay we need to
collaborate and oh right
salespeople talking to engineering
into production people talking but that
hasn't happen there different tribes
they have different language yeah
there's a lovely one my favorite Swedish
phrases when you're talking to farmers
use farmers words yeah and the whole
version is and when you're speaking to
students speak Latin and then Toby the
farmers use farmer's word so when I'm
talking to accountants I should be
software engineer and I'm agile so I
talk about story points and sprints and
burn up and burn down and they go yeah
right yeah right show me the money
like oh no if I'm delivering value so we
need to collaborate across divisions
across silos and the weird thing is all
parts of the organization are seen as
value creation because they are once I
can start connecting all the pieces up I
enabled some business capability over
here which made us some money or
increased customers or reduced churn or
did something move some business dial
and because I can equate now I can
associate the work with the value I've
now got two levers I've got the cost
lever and the value lever okay so so
what so this then sorta led me to my
second insight which is rules for coping
aren't just policy rules for coping
become law okay
there's a set of practices called GARP
generally accepted accounting practices
and got is how businesses how listed
companies have to do their math it's how
they have to present their results to
the market so that all of the companies
can be measured equally and Garf is
entirely based on cost accounting
principles and so that means that GARP
like the law literally the law says I
cannot use throughput accounting unless
I also jump through some hoops to make
up the equivalent cost accounting
figures all right so just Dez off
famously just through that of in he said
that if you want a call to these
shareholder blah blah blah go invest in
something else cuz I'm obsessing about
customers and I'm obsessing about making
opened up a part of his organization's
internal acquisition of servers and
things like how I as an Amazon team get
a server and he
opened that up for the world and called
it Amazon Web Services and created a
billion dollar industry overnight smart
lad yeah so because he's obsessing about
the value chain and he's obsessing out
than cost accounting loads and loads of
other companies are trying to replicate
this and they can't because they're
stuck in the existing rules that keeping
it in place oh dear
so then is well one of the most fun
things about putting this talk together
was finding word clouds I challenge you
to Google for cloud word cloud and find
a word cloud about cloud in less than
about two hours because this really hard
anyway here is here is a slightly
doctored word cloud about cloud so cloud
my first real encounter with and I guess
really its virtualization so much as
cloud so cloud is probably the most
significant shift in IT I think in
technology in the last decade it flipped
the whole model on its head I remember I
was working at a a well-known bank in
the UK in 2007 so I was still at
footwork then and and they had like they
were one of the early adopters of VM so
you had this VM farms and they were
they're repurposing some of their old
mainframes as VMs which is pretty cool
and and so we said are great so and
we're going to need to get a VM and they
said oh you're going to need to fill in
a form for that so what kind of form it
was a procurement form for a server like
like the server's already there what we
want is for you to just let us offer on
ah procurement it's a fervor of like
this is really silly it wasn't as silly
as once we got our server with our one
gig of memory and I wanted to have two
gigs of memory and I went ask for human
the memory is already in the machine you
need to turn it up no no you're gonna
filled in your TPS sheets right it was
clearly bonkers yeah mmm so there's
interesting that there's a number of
theories around whether or who who can
claim to have invented
cloud computing or first proposed it
there's a chap called Linklater
who was one of the original darf a team
that invented the ARPANET and he was
talking about in a networked from
so that you didn't really know where
they were we didn't have to know where
inventor of Lisp and a whole bunch of
other things one of my favorite
observations recently there's a guy
called Thomas J Watson who was the
original CEO of IBM so he was a nineteen
fourteen he took over IBM he was going
in 1940 it wasn't called IBM then he
renamed it to International Business
Machines and he was a CEO there until
like the 50s actually died in 1957 and
he famously said sometime in the forties
than five or six computers I can
envisage that the world will never see
more and I always been you know one of
those schoolboy schoolgirl kind of
chuckling moments ago I wanted computers
I only look around and you've got Amazon
Asia Google bluemix right we're probably
going to end up with five or six
computers they're just really big
century ahead of his time
this multi-core thing multi-core can
supercomputer beast in my pocket is
merely an access portal to the real
computers we are going to end up with
about five or six computers so yeah this
has been weird trying to get dissing on
some kind of cloud you know public cloud
transition thing in their work at the
moment right keep your hand off as it's
going well Wow if you haven't stayed up
fantastic that's great most got most
organizations are really struggling with
this so let's take a look at this then
well the first thing is clearly is on
demand compute power right that's the
first thing I want can i if this is the
power that cloud gives me is on demand
computing that's never existed before
so what limitation does that diminish
this is a little bit more subtle the
limitation it diminishes is the cost and
risk of running your own data center now
this is a trade-off because there's a
lot of upside to running your own data
center as well if you're dealing with
latest latency sensitive things and
whatever else or high security things
then you may well want your own data
center but it's very typically very
costly very risky to run your own data
center and also you've got physical 10
and physical 10 days
by physicality by buying and selling and
shifting and moving things around if
it's someone else's 10 its virtualized
completely and it's public cloud I have
a lot more options so what were the
rules then that were enabling us to deal
with this limitation with this without
public cloud without with our data
center limitation which things like
procurement maintenance decommissioning
the whole lifecycle of hardware the
whole the whole management of Hardware
all of those things are expensive that's
one of the rules right be aware that
hardware is expensive don't just go
chucking it around so you okay them or
that mean computer hardware requires
lots of people to look after it lots of
different kinds of people so you've got
data center engineers networking
infrastructure security there's a whole
raft of different people and again
remember we're still in our cost
accounted welds are they're in silos and
they don't collaborate very much because
see previous slide and so we need a lot
of people okay to look after it and all
around hardware which we think of a
capex capital expenditure writes money
it's amortized value and all that
they're expensive and they require lots
of governance okay so lots of ceremony
and that's the world we live in and so
here we go well let's assume none of
that was true
anymore if we unpack all those rules
what kind of things I'm really need to
speed up a lot what kind of things are
we are going to help us then work in
this new world well the first thing the
first new rule is I can explore ideas
really quickly really inexpensively ok
all of our processes in most
organizations are not set up to support
that kind of working there's a huge
shift at the moment lots of
organizations are on this
transformational journey or whatever
they might call it to try and suck less
at doing small things quickly yeah
here's a mad idea once I've bought
something I can switch it off right I
can reduce this compute power as easily
as I can increase it it's a dial it's
literally utility it's insane and so all
of my thinking about because part of
this is optionality so I'm thinking
options I bought this thing
my options now I sell it at a massive
loss right or keep it and try and force
people to use it and add more time
whatever the cost of use but if I can
reduce computer power as easily as
increasing it means that those decisions
are now much more reversible the cost of
making a bad decision is much lower the
cost of reversing that decision is much
lower I'm renting compute power that's
my mental model I'm renting to do power
are not buying service and that's a
whole different lot of thinking that we
do so now again mm-hmm this is where
these rules start really impacting
things because now they become structure
they become how the organization works
so we've got lots of people and these
people are specialists ok and they're
expensive and we have governance and
ceremony and all that kind of thing
these people I'm not going to say to
cede their power quickly so this is in
direct conflict with this idea of
reducing compute power yeah well if it's
hi ceremony I don't want it to be easy
to undo these things I don't want to
believe it's easy to undo these things
and then finally very briefly continuous
delivery so a lot of people here I hope
of practicing continuous delivery are on
their way there so again what does it
give us it gives us the ability to
eliminate simplify automates releasing
software by the way you should do it in
that order so I look at my process and I
say what parts of this process can I
eliminate what parts of vestigial then
given what's left that I have to do how
can I simplify that and finally then
what of that simplified process can I
automate so that's the offer that's the
deal okay so what limitation does that
diminish it means I don't have high risk
releases it means I don't have tried
transaction cost releases anymore which
means I can release smaller things
that's kind of cool
I reduces bottlenecks specialist so I
have all these gates that I have sign
offs of what's it called change board
cabs and I have various things that have
the word Authority after them so design
authorities and security authorities and
and a bunch of other high status people
who walk around the building and
typically in suits saying no none shall
pass so the rules I currently have of
this fixing mistakes will be expensive
and time-consuming so I front-load the
risk if I don't front like
the risk bad things are going to happen
and we've all lived bad things okay
we've all been there or there do the
more gnarly of you at the tie-in over
the crow's feet around your eyes have
done those two three four a.m. Saturday
night releases right where it didn't go
right the first three times and you
couldn't eat any more pizza
okay so fixating that's going to be
expensive and time-consuming managing
the risk then requires specialists to
check things manually because that's
what we do we pour through these things
so what new rules am I going to need
well I'm going to need to entrain
specialist knowledge rather than having
it trapped in the heads of specialists
what I want is I want self evidencing
processes in other words by doing this
it has to be secured by doing this it
has to be compliant okay look up as a
fun thing I was 18 F which is an
organization in the state that was
assembled very quickly to fix the disaster billions of
dollars health care system could deal
with about 14 concurrent users rather
than 320 million and then a bunch of
jokers rebuilt the whole thing in about
eight weeks because it's a website and
and what they've figured on the way
through was the they've really waded
different compliance things and boxes I
had to check and they managed to
automate almost all of it
it's a great great story so you create
them make the process self evidencing by
the time I come to release I'm already
compliant secure scalable all those
other things I can't not be I can't not
have got this stage and B it means like
in slice work into small valuable
increments right the transaction cost
comes down I can release things smaller
I can release things more valuable I can
get feedback faster it's all win here's
a fun thing everyone needs to learn how
to automate right we're all programmers
now yeah so everyone in the organization
is thinking how can I simplify eliminate
automate the work I need to do so now
the rules become culture and this is
where it gets really really tough to
unpick these things okay if we believe
and act as though fixing mistakes is
expensive and time-consuming and these
specialists are somehow elevated holy
people and then we're saying well
actually anyone can pick this stuff up
go learn Python you'll be fine right and
by the way we've interested
we made tools that mean that most of
your specialist knowledge is now
enshrined in those tools so we need you
to make sure the tools are okay but you
don't get to do the nun shell fasting
anymore you get to be advisors rather
than gatekeepers and some people are
like yeah we can get work done and a lot
of people like no I lose my status okay
so then what can we do how can we break
the rules let's go through this quickly
understand the power of a new technology
so all I've done is I've taken gold rats
for questions when I flip them into
statement okay he's asking what is the
power of the internal urges we need to
understand it so we need to ask some
questions what does it do how does it
think of this as a checklist right will
it work for us is this going to work the
same way for us or is this just blurb
and how can we exploit this technology
it's cool but is it actually useful is
it going to be valuable in our world or
is it just interesting looking and then
we want to recognize the limitation that
the technology will diminish because if
we can't see that we can't know whether
the things working yeah so how can I
prove that the limitation is holding me
back what could you do what experiment
could you set up to prove that the
limitations holding you back is it just
a what's it called a limiting belief
right and so for instance one
organization I'm working in absolutely
convinced that they need to increase the
size of the teams that they have because
their problem is capacity and then we
did a bit of work with a value stream
mapping and it turns out there's three
significant pieces of work that are
backed up at release because the past
alive is really really bumpy and manual
and error-prone and so each of those
things has to go through one at a time
slowly and so if we doubled capacity for
development we now have six things
waiting unable to be released yeah
rather than ramping up a small army
let's take some of those people and
point them at the past alive problem
let's understand and this is Theory of
Constraints again it's understand where
the system is constrained and wider than
that constraint and so that limitation
of the ability of being able to deliver
stuff faster got our great let's let's
adopt this technology it's not going to
help us we managed to prove that that
limitation it is a limited
it will eventually be a limitation but
right now it's not the narrows part of
the pipe how would you know if we
adopted this technology that that
limitation was diminishing how would you
know things are getting better
what could you measure so let's get into
metrics and flow and some of those kind
of things right how what what
experiments can we set up to say let's
adopt this technology let's adopt this
idea and see whether it helps us and
what our control group maybe it's just a
fluke right maybe we're going to try
this experiment and then do you know
what happened things got better
yeah there's there's I company of the
guy's name it'll come to me in a second
but a whole series of experiments in the
nineteen twenties and a factory you
remember the factory and it was a the
Hawthorne Thank You Hawthorne effect
this chap Hawthorne and he was in the
saturated measuring thing he said right
so let's figure out how we can get
people in this factory to work better
and so they tried dropping the
temperature by a couple of degrees and
people's work rate increase so they went
all making the room colder people work
faster and then but they noticed it
tailed off no no and so they try making
it warmer a couple of degrees and people
work fast women
oh wait wall the rooms make people work
faster oh and then it's held off and
then it said what if when you change the
lighting and the light emits slightly
dimmer and people work faster that our
people work better in low light what was
actually happening was this every time
you mucked around with the environment
the people there going are they're
watching again oh it's the bosses up in
the galley oh we're on it old areas ha
ha better keep at it right think you
can't that's not sustainable so
eventually you slow down and go back to
so this thing where you muck around with
the system and you think you've made an
improvement yeah
what's your control group do you have a
control group if not how can you counter
for some of those sampling biases and
other measurement biases sophistical
biases we need to identify the existing
rules these to manage the limitations
this is generally really hard is what
I'll say because they're hiding in plain
sight they're just what we do the word
just appears a lot when you're looking
for these limit paper we just do that
just anyone does that why would you not
do that well because right so how will
those rules get in the way so what
assumptions are those rules making that
are true of the existing system that the
new technology
going to break those assumptions how do
we challenge those assumptions in
particular let's look at ownership now
now I'm gonna start like indicate grade
C I'm going to start saying who might be
threatened by dismantling those existing
structures because those structures have
people and people have status and
peoples of status is emotionally
attached to some of that stuff if we're
going to dismantle that how do we do
that in a safe way how do we find our
our strong and weak support and how do
you find our strongly composition and
how do we start working with those
different communities to to make sense
of that if you don't understand what I
mean by strong and weak support or
strong a weak opposition then come on to
Kate gray and Chris Young's talk
tomorrow he says plugging it which I
suspect is going to look at that and
it's a very very powerful model how can
we make it safe to change right if
someone wants to come on the journey how
do we help them on the journey if they
don't we need to give them a graceful
exit how do we make it okay to say you
know that X million quid that I felt
thought was going to X million euros
that I thought was going to be great to
spend on that thing
not so much yeah but now I lose face by
doing that so how do we say well that
was a great idea then let's see what's
beyond that how do we create graceful
exits for people and finally how do we
identify and implement the new rules and
this is where all of Linda's fearless
change stuff starts to appear right so
how do we safely exploit this new
technology I want to use this thing but
it needs to be safe and there's a
community safety thing as well as a
technical safety thing how do we make it
safe for the organization like Maslow's
hierarchy of needs kind of safe to try
these things out and what
contraindication should I be looking for
right what things what unexpected side
effects might I have because if I'm
looking at where I expect it to be this
is my confirmation bias again I'm not
looking over here at this thing blowing
up and even if I notice it blowing up at
the corner of my eye it's not me it's
this and this thing happens to blow up
could it happen to anyone
right and so finally then how do we
interested you introduce and how do we
institutionalize these new rules because
we want to make these things normal
who's going to own these new rules okay
so there are some ways to break the
rules some things to think about the
breaking the rules what's the new
technology what power does it have
what limitation does it diminish what
rules are we going to need to dismantle
and what new rules are we going to need
to introduce an institutionalize if we
can go through that sequence we have a
fighting chance of really exploiting
that new technology if we don't it will
fail and will blame the technology
that's what we usually do I actual fact
we blame the people who brought us the
technology we kick them out because
we're a tribe and we close ranks yeah so
there's an awful lot to this statement
technology can bring benefit if and only
if it diminishes a limitation I would
say very much I would encourage those
who've read the goal to go and listen to
beyond the goal is wonderful wonderful
series of lectures but if you haven't
read the goal start there it's a
fantastic story in the meantime don't
break some rules thank you