Okay, our next speaker is Amanda Cassini.
We are very fortunate to be able to get a talk in the open hardware dev room on one
of the topics that in the open source community is much more easily addressed in software,
but is harder in hardware as some things are.
So please give a warm welcome to Amanda Cassini.
Thank you so much for inviting me here today and for inviting me to speak.
I'm very excited to talk to all of you about economics and economic modeling, which to
be quite frank is actually not my background, but is something that is very important for
the work I do now.
And so who am I in case you're wondering?
So I'm Amanda Cassari.
I'm a pale white woman with light hair and eyes.
I usually wear glasses, which I should put on now if I want to keep reading things.
I'm a researcher and engineer at Google and I'm currently leading a team focused on research
and education in our open source programs office.
I'm also a co-lead for Project Ocean and Across, which are an external faculty to the
Vermont Complex Systems Center and a co-founder of Open Source Stories with Julia Farrioli.
Who's here today?
Thanks for being here, Julia.
I sit on the board of directors for something called the Computational Democracy Project
and I once wrote a book with Alice Zhang on feature engineering.
I'm also queer, a very proud mom of two smaller humans, a US Navy veteran and lucky enough
to live in the indomitable state of Vermont in the United States.
It's just south of Montreal, which is easier for some people to know.
And if it helps you better to follow along or understand what we're working on right
now or what these slides are, please, there's a bit.ly that you can see all of these slides
and my speaker notes, which will amount to a transcript.
And that's bit.ly backslash e-c-o-n-o-m-i-c-s dash o-f dash o-s-s dash dash f-o-s-d-e-m two
zero two four.
And so I also just want to put in the caveat.
I'm very sorry, but something about listening to things from 1.2 to 1.5 speed means I usually
talk at that speed now as well.
So hopefully if you miss something, you can go back and look at the transcript and speakers
later.
And I'll just make sure I'm being at a good accessible rate.
So like I said, when I gave my introduction, you may have noticed I am not an economist.
However, I am a complexity scientist and I try to see the world through a mix of applied
mathematical models and abstractions.
And it's this background that I'll be leaning into as we explore economics of open source
and open hardware that exists right now.
I guess to a little bit more just background, why do I care about this?
So I've been working on this.
In 2019, I was originally working in Google Cloud as a DevRel manager and an amazing team,
but I also had this really nagging feeling that there was something kind of understanding
about the way that technology and open source worked that we were fundamentally missing
out on and we were trying to estimate and understand our work.
And that's how I met actually through Julia, was the half bakery of ideas and discussing
something.
I was like, I have a niggling feeling of something we should be working on.
She agreed and together we were able to pitch an idea for what is called Project Ocean.
And that's a pilot.
It was a pilot, it's continued on now and that's looking at collaboration between academia,
industry and communities to understand open source at a global scale.
We did launch this in early 2020, rough time to start a new project, but we were able to
achieve many of the goals that we had then and that work does continue in partnerships
and collaborations now.
One of the most audacious goals that we had was mapping out the entire open source ecosystem
and then sharing that with everybody.
And it sounded honestly at the time like it was not that audacious because it just sounded
like it was a way of looking at information and then sharing that in a way that everybody
could see, which was something that we all deal normal basis of just making information
transparent.
But one of the barriers that we ran into then and we continue to run into now is just even
defining the problem space of open source with stakeholders and communities as a shared
understanding.
If you've ever done any kind of research, that's always the place you have to start
as being able to actually define the boundaries and the constraints of the problems that you're
working within.
And the reason that this was a challenge wasn't because there wasn't precedence in explanations
of how you could go about that.
It's that we were finding that a lot of the mental models and analogies that were used
were not actually universal or global.
Those models were actually preventing us from having a deeper understanding of the complexity
of what was happening in open source because they were too simplistic or honestly in some
caries is just the analogy that no longer existed today and was not the reality of the
world that we were living in.
So that was really hard for us because all of these models also made an assumption about
the kind of baseline that you're working off of.
So it gave you an understanding of like, well, this is what we should all assume in terms
of the number of people involved, the kind of work that is critical, the amount of money
that is existing and transacting to be able to keep this ecosystem thriving.
And without all of that, this is where problems might be existing.
So if we were looking at this concept of risk and resiliency, you need that baseline to
understand where risk and resiliency either has interruptions or needs additional resourcing.
So we were struggling with this fact mostly because organizationally, as a group and within
this larger company, it was challenging for us to understand how we could move forward
not just effectively but in a way that moved with purpose.
And we can't do that until we start breaking some of those popular over-reductionist models
of how digital structure looks today.
This is a very popular comic.
You have not seen this yet.
This is used a lot to show demonstration of gaps, not only exist in understanding but
also in a collective failure to respond to the challenge that exists for this large global
ecosystem.
So just to describe it briefly, this is the XKCD comic.
There's a bunch of blocks.
There's this idea that this represents all of digital infrastructure.
And at the very bottom in a critical space is a person from Nebraska who has been thanklessly
maintaining something since 2003.
And more often than not, I mean this comic gets thrown around quite a bit, but I've
actually seen this in position papers as a reason for why someone should be getting
billions of dollars to do research or initiatives.
This is their example they're giving.
It's supposed to be demonstrative of the fact that the open source ecosystem is brittle
and that large-scale investment is critical.
But for that very specific purpose that that group wants to get billions of dollars for.
And this is why this is my counter to this comic, is that the reality is that we're already
spending billions of dollars on sustaining open source.
I dislike the term sustaining because you can sustain broken systems.
It doesn't mean it's working or it's working well.
It just means it can keep going.
And so because this has been so effective in centralizing buckets of money and attention,
there are walls that are being built around organizational spaces for collaboration that
we have to break through in order to keep moving forward.
It's not that we're not investing, it's not that investments are happening, but it is
that increasingly we are localizing decisions for investments and support in a way that
is not creating resilient systems.
Which is a problem when the system you're dealing with is not localized, does not have
local optimans.
It's in fact global, decentralized, and it's an equitably resourced.
So to move forward we need new approaches for both challenging the assumptions as well
as providing well at paths as understanding how we can approach this differently.
So this brings us to the impetus behind mapping out these open source ecosystems.
So not just creating this natural understanding, but really what this whole idea was about
is that we were assuming and we still assume that if we have better information and better
ways to understand the world we can make better decisions.
And this is important for my work especially and what I get paid to do because that's
understanding where resources that we have available should go.
And they always have to be in the context of the business.
That's what working for a for-profit company constrains me to.
But we have to know where we are so we can work collectively to understand where we
need to go next.
And one way of knowing where we are is to always look at where we come from.
And I don't want to discourage the idea that history cannot teach us what we should be
working on and where to go next.
However, some of the work we've defined as black swans, these are events that have fun
and minimally changed and impacted how we approach open source today.
They also sometimes get used as a scare tactic.
And again to centralize over centralized resources in an already constrained environment
and sometimes by the very organizations that we are trusting to work in our larger best
interest.
It may be true that there is an intolerable amount of risk that exists not only because
of incidents like these but that we haven't even seen yet.
And it might be that investing in those in a way now will prevent something further in
the future.
It might also be that the next problem that happens like this will also only be visible
through hindsight and no amount of centralization right now will prevent that.
So we can be informed by the past but we shouldn't be scared by it.
So I think that my argument here is that we should be able to move forward with a critical
eye again with like looking at better information, better understanding of what currently exists
now.
I've previously talked about tools and frameworks that organizations like open source programs
offices can use to identify stakeholders and especially in companies as well as how to
define and design metrics for regular reporting.
These are honestly frameworks that are useful for any type of work not just open source
but they do help in moving forward with something that's very large and messy where you don't
have an ability to look at customers coming in and things going out.
And those are building blocks of information for teams who work in an open source to frame
the value of our work for businesses, partners, communities.
This work was not created either randomly or on a whim or in a vacuum.
So our team together had been working towards this very specifically because we were asked
at the time to develop an ROI model of open source that we could use across the board,
across the business.
And this was for all of Google, for all the work that we did, for all the investments
we put in.
How can we describe that using an ROI model?
My problem at the time which still exists and exists now is that I still maintain that
ROI is almost always the wrong economic model to use for open source at scale.
The reason for that is ROI is a very specific economic model that assumes first order inputs
and outputs.
And so those first order inputs and outputs mean you always know what's going in.
You can see those direct effects and it's always an output.
If you are a network scientist, this means you are always looking at a first order system.
Open source is not a first order system.
And so if we try to boil it down and make it a first order system, we are actually either
abstracting away or making an overreductionist model that does not actually serve us.
One example of this is when projects are tried to be measured simply by lines of code.
Lines of code has not actually been shown over time to be something that is an indication
of either productivity or even efficiency.
When it is used to judge productivity and creativity, then we run into large problems
with teams being devalued.
We can't simplify that away.
So what are the abstracts we should be using in taxonomies?
I realize I'm going to go through this fairly quickly.
So all of this is grounded in the idea, by the way, that the current structures and systems
we're working within are the ones we have to work within.
There's more research that goes through these problems and instructions using critical theory.
I'll be taking a pragmatist approach today and just be talking about it in terms of the
world we're currently living in, trying to describe those a little bit better.
So resources.
Resources in this capacity, I'm talking about an abstraction of not just money.
And the reason for that, again, is that we boil things down to money quite a bit when
we want to understand what kind of resources we're looking for and what resources are important
to us.
So there's just a few listed here.
And again, these are somewhat abstract.
This does not even account for things.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
Thank you.
For just playing the
call.
This was actually,
I think this was an interesting
achievement though.
It also means that if you were considering these kinds of fundings you should understand
You
You
You
That's called op-ex operational expenses and that changes year-over-year for companies and organizations
Academic institutions always have to take into account some kind of general institutional fee
So if you are working as part of an academic university or an academic project group and they get a grant
You're not going to get all of that money
So there has to be costs associated with it
You need to take that into account again as part of your budgeting and as that being your fiscal host
Another part is there's a lot of government funds and NGO funds that are starting up right now
Those may be restricted only to citizens of that government and then we run into the problem of how do you talk about?
origination and citizens for a global project
All monetary funding requires fiscal hosts we went over that
LLC a nonprofit an organization with a platform and the reason that exists
It may feel like that's blocking you the reason that actually exists is mostly related to anti bribery laws
Organizations and companies can't just hand people money and then have no record of it have no tax record of it
These all basically funded the same kind of guidance as being able to identify and talk about things
Okay, so quickly outputs versus outcomes. So again before talking about these larger modeling structures
I feel like it's important to differentiate these two
Outputs of your project are the very specific units of work which you've identified in your roadmap or plan
This is the how of what you want to achieve
Outcomes are slightly different. They address the so what of your roadmap and your proposals
So this is how you identify the purpose of what you're doing
It allows you to frame the importance of your work in a way that may not be readily available to a general community or the group of
Funders that you're trying to target
So when you were tracking your reporting what comes from your work, you want to be able to address both of these things outcomes and outputs
So very specific for open hardware. I'm so sorry. It took me so long to get here
The original creator of the bomb
Right so part of bills of materials and challenges that comes with funding on bills of materials is this question of centralization and
Decentralization of hardware component availability
So as you're trying to look and plan out and explain your project things along the way
There's a big difference between when you first write your design proposal and then down the line
Maybe when you get funding then when you can act on that and when you have to deliver on outcomes and out
Outcomes and outputs. What is the component availability that you have and has that changed you to other market forces?
high maintainer costs again inequitable
Inequitable availability of parts
It's difficult always to find people who may have the specific skills that you need and when you're operating on these tight timelines
If you do not have a robust group that is working together and you need to do some kinds of swaps
Or you need to be able to have someone who can maintain a changing bomb over time
Then you're gonna have to make sure that you are
Working on those in a way where you're not necessarily going to be able to depend on depend a bot
You may not be aware that something was either obsolete or that was impacted on a timeline that now moves the entire project weeks at a time
So get lead and delivery times also the challenge of one versus ten versus ten thousand when you're trying to order parts
And when you're trying to get things from manufacturers, they have to be able to create and work on those in batches
Also, your household may not be have the size to be able to have
30,000 of one tiny piece in a box that you only need 10 of
So just in terms of funding asking for funding when you're thinking about how to ask for different parts
That's also something you have to consider based on the kinds of project you're building or what other people are building is
Going to be as well that consideration of how do you share resources that are people or time?
Okay, there's no option up not every single kind of development has an open option for all kinds of hardware
So especially working with different kinds of chips different kind of microprocessors sometimes the only
Option that you may have is proprietary software to be able to work with your hardware
You have to be able to budget for that as well
And that's difficult for folks to understand who are used to funding and all of the tool chain
They're working with is an entirely open tool chain that doesn't have a cost
Associated with it licensing fees are absolutely something that you should be including as part of your budget
You have the option to build it yourself, but you may not always have the option to build it yourself
So working with proprietary tools puts an additional cost on your budgets
Working on funding lack of funding. There's a lot of conversations around software funding
There's not as many around hardware funding at least not for specific purposes
So I think in general this is also a challenge when just when working between different kinds of funders and funding
Where is it that you fit that you feel comfortable with?
There's a higher cost of getting it right the first time when you have to order something
Then there's a long lead time and then it shows up and you realize that you've crossed a few of your paths
And now things blow up
But software you can just go through and re-change it and then redeploy it
It doesn't exist the same when you're trying to connect components for hardware
So okay very quickly. I'm so sorry
ROI ROI actually can be used sometimes
So I don't agree with it as a large-scale
Model to work with however if you are working on something like a contract or you're working on a very specific outcome of work
This is a good time to use ROI in your proposals
So being able again those identify those inputs the kinds of work that you're doing that out comes in the outputs
This is something that you can boil down to something like money and time
As long as you're always taking into account things that do have monetary equivalents and direct effects
There's also something called XYZ model
This is basically assuming a black box in the middle
So the red is kind of inputs the blue is outputs you have a feedback loop
What happens in between is a little bit fuzzy, but because you're able to describe it and again from that like kind of outcome perspective
Then it's easier for people to understand what you're doing and why
Another kind this actually recently came and it's a good example of this from
Hoffman at all paper the value of open source they take a supply and demand model
They worked with some very specific people to understand what kind of things they're using within the corporate industry
And that allows them to look at things more from a supply and demand perspective
So there's even more things especially that apply to software and hardware
There's many other microeconomic models that I think that can be applied and have be tried in this case
Again, I think the main takeaway hopefully for this and the challenges that continue to exist is that there's not one that fits all
There's not one that fits every project
But a lot of funding is designed to examine things through one very specific project being clear through as you walk through and
Identifying your needs as well as identifying what resources you have and what the outcomes will be will depend on your project
Don't try to squeeze it into the model
They're asking for if you feel that that's not going to accurately represent the needs that you have all it will do will cause
What's exactly happening to somebody I know right now which is they have a week left and a bent frame with a four week time frame
Or a four week lead time
So they tried to save some money
Didn't work out materially and now they're going to be spending six times as much to get to that outcome
Part of that was through inability to look at resource planning and explaining things in a way that actually helps there
The person who they're working on identify it as something that they should be spending that money up front on
So collectively understand where we're not on the same boat. We're in the same storm
We're not on the same boat, but hopefully our flotilla together
Is a place where we can continue working to support each other and where we want to be
Thank you
You