should be at least a little bit harder
to read and I think that we’d all be
maybe a lot better off and I’ll try to
convince you but first I’d like to tell
you just a little bit about myself as
you’ve heard I’m passionate about games
I’m passionate about computer
programming and I’m passionate about the
law
what are those three things have in
common rules I’m a guy who likes rules
not rules for their own sake but what
rules are capable of creating like the
rules of Go Go has eight rules give or
take you can express them in about 102
English words if it’s on a business card
the tiny set of rules has generated so
far three thousand years of people
playing and winning and losing and
teaching and learning and growing and
fun I think that the rules of go are a
masterpiece of rule writing I like rules
so I like computer programming which is
essentially just giving rules to a rule
following machine and eventually my love
of rules brought me to law school where
I was a wash in rules rules about how to
write rules rules about how to read
rules and rules about when other rules
should be overruled you’d think it had
been great but there was a problem there
is a problem because I learned computer
programming before I learned the law and
now when I read Lytton written legal
rules I get this uncomfortable feeling I
get this feeling that there’s a missed
opportunity here and I like to share
that opportunity with you but first if I
could I’d like to try and share that
who’s winning this game of poker all
right you can stop it’s the hand on the
left that’s not important what’s
important is while you were trying to
figure it out did you have that little
voice in the back of your head that was
telling you this is why is he making
this harder than this needs to be
this is sub-optimal there is a better
way to do playing cards right that’s the
way that a computer programmer feels
whenever they read a law or a contract
why because computer programmers learn
multiple languages they learned that
languages are tools and you should
choose your language based on the task
in front of you and for laws and for
contracts that are written in natural
languages like English in French it’s
just really obvious that this is the
wrong tool for the job let me show you
why with a little math quiz ready what
is 3 plus 4 multiplied by 2 come on okay
I heard some 14s totally different
question what is 3 plus 4 multiplied by
2 11 I said the exact same words twice
now what’s interesting to me here is
that we can all agree because we know
the syntax and the semantics of math
that the right answer for the second
question is 11 what’s interesting to me
about the first question is not what the
right answer is but whether we can
realistically say that it has won
everybody said 14 so does it have a
right answer is 14 a good answer and 11
a good answer yeah I think so I think
they’re both reasonable interpretations
of that statement which is great if what
you’re trying to write is a kind of
mathematical poetry I guess but if the
difference between 14 and 11 is
important than a natural language is not
the right tool for the job
the great thing about natural languages
is that they allow you to say literally
anything the great thing about
constrained languages like computer
programming languages or like math is
that when you say anything it can only
mean one thing and law like math and
like computing computer programming is
an area where knowing exactly what the
words mean is really valuable so this
sense of frustration that I had
eventually led me back to the University
where I’m now a student in computational
law and part of my research is looking
at languages other than natural
languages that are good for expressing
legal concepts and what it’s possible to
do with them when I say that laws should
be harder to read I mean they should
look more like this isn’t that beautiful
doesn’t it look like a typographical
yardsale it’s a mess but it’s wonderful
it’s a programming language called ergo
that was designed specifically for
expressing legal concepts and I think I
think it’s the greatest thing since
sliced bread I’ll explain why this like
the rules of go is a masterpiece it was
originally expressed with wood and paint
there’s no wood and paint here today the
original masterpiece has been translated
it’s been translated into a language of
color codes and pixels and that language
has been written down in a computer file
that’s on a computer or there somewhere
I think I can’t read
that language and neither can you we
can’t read JPEG but we don’t need to the
computer can read it and the projector
can display it and we all don’t have to
go to France to see it what I’m talking
about here is not revolutionary its
digitizing information taking
information and turning it into a format
that a computer can read and write and
can use to do useful things we’ve done
it for pictures we’ve done it for video
we’ve done it for audio we’ve done it
for documents and what used to be called
because the law isn’t about documents
the laws about rules
we haven’t digitized rules I think the
reason why not is that people just
aren’t aware what you can do so let me
give you an example of that as a lawyer
my client will come to me and they’ll
say Jason I want this legal outcome I
want to comply with this law give me
advice tell me what to do
so the client has the legal outcome they
want I’m supposed to know what the law
is and then I take those two and I
calculate out what the person should do
what facts should we create so that if
and when the horrible thing happens you
end up in court you’re going to be found
in compliance with that law I wanted to
see if I could use this programming
language to get computers to give legal
advice in the same way now to be certain
that it was giving good legal advice I
went and I took a famous Supreme Court
intellectual property case and I took
those three elements from that case the
legal outcome of that case I imagined
okay this is what my client is asking me
to ensure happens so I took that legal
outcome and I encoded it in the
programming language and then I took the
law that was referred to in that case or
at least a subset of it and I encoded
that into the programming language but I
didn’t want to just tell my software
what the answer was so I
wrote the software so that it would
understand a wide array of possible fact
scenarios across six different variables
and then I asked the software what
should my client do what is your advice
and I thought to myself this experiment
will be successful if the computer tells
me that what my client should do is the
same thing that the people had actually
done in that court case makes sense
right so this is what it actually looked
like again it’s not pretty it’s not easy
but the important thing to know is that
I didn’t tell the computer how to figure
this out all that I did was translate
the law into a programming language and
translate the desired legal income into
a programming language and give the
computer a framework of possible fact
scenarios so that it could generate
random ones I didn’t tell it how to do
the mental math involved
I just translated so did it give me the
right answer yes and no it gave me three
options so initially I thought oh I’ve
I’ve felt this up but I checked and one
of those three options was exactly what
had happened in that court case so then
I checked the other two options and they
were right – there were three ways
within those six variables of getting
that legal outcome I was only aware of
the one I was just hoping it would find
the one but it found all three that
existed out of hundreds of possibilities
could I have done that for my client yes
it would have taken me at least a few
hours it took the laptop 30 seconds so
what can you do if you digitize rules
well you can tell the software what you
know about the law and then the software
can tell you what you don’t know about
the law you can automate legal advice
giving
which just suggests a huge huge range of
possible applications I think the best
way to understand where the legal
profession is right now is by comparison
to the profession of accounting it’s
possible but hard for an accountant to
use a computer programming language to
get the computer to do math for them and
like my experiment demonstrated it’s
possible but hard for a lawyer to get a
computer programming language to get a
computer to do legal reasoning for them
spreadsheets came along and all of a
sudden it was easy for accountants to
get a computer to do math some analogous
thing that I call the killer legal app
is coming and it is going to make it
easy for computer for lawyers to get
computers to do legal reasoning when
that happens the history from accounting
tells us that everything changes the
cost of most if not all legal services
will go down for some legal services
it’ll go down to zero new kinds of legal
services that didn’t exist before
because they would have required the
lawyer to sit thinking for too long
those kinds of legal services will come
into existence and far from the robots
are coming for our jobs fear the demand
for lawyers is actually going to go up
because lawyers are going to be able to
do more they’re going to be able to do
it more efficiently and they’re going to
be able to do it more accurately that’s
what happened in accounting now the
killer legal app isn’t here yet there’s
some people working on it but it’s not
here yet it’s coming and when it comes
we need to digitize laws for it to work
as long as we are taking laws that are
written in a unrestrained natural
language that process of digitizing them
is going to be needlessly difficult like
making laws harder to read is not a
thing in and of itself that we want
making laws harder to read is the price
of something that we want it’s the price
of finally bringing the legal profession
into the digital era and I think it’s a
price well worth paying thank you [Applause]