The Strategy Rant

[decentralion]

Today I wrote a long “strategy rant” in the Discord. At @LB’s suggestion, I’m posting it here to keep track of it, so I can later try to work it into a doc.

It started when @deltafreq asked in Discord:

@decentralion my understanding of the current business strategy is 1) harden product in beta, 2) move into crypto, 3) move into open source. I totally see the logic here. Crypto is liquid, fast growing, and full of early adopters. Open source already has natural engineering synergies with SC and integration solves for them a long standing and well understood pain point of contributor compensation. I think this will work. I do want to point out a couple observations about it that just popped into my head though. A) from a macro-economic perspective, the entire crypo market-cap is tiny. All cryptocurrencies added together are worth <= a single FAANG company. I assume similar is true for open source. This isn’t necessarily a problem, there is no rush I suppose, but it is clear that landing a single big corporate client post beta could pay all the bills many times over. Just something to consider.

Then I was thinking about what is the value prop to a CEO and I think this is where SC really shines. Automated value attribution and compensation isn’t some niche application, it’s a fundamental technology that I think will eventually be a component of most companies. Right now HR has to do this manually and I imagine that in really large organizations its painfully inneficient. What is employee A really worth? You help them answer that question. But its more than that, because not only do you map their value contribution, but you provide employee A a map of how best to add value (we were talking about this with @LBS and @rpp). So when all is said and done, that CEO can potentially fire a large chuck of his HR department AND project managers… It’s uncomfortable, but that is the level of efficiency that I think SC could create… it might really be that disruptive.

The reply:

so, @DeltaFreq, you’re right on all counts: - good summary of SC’s near and medium term trajectory and why crypto makes sense as early adopters - crypto is actually a small market, and if we were looking to maximize near-term revenues, we should focus on getting tech companies onboard moreso than open source, since that’s where the money is - sourcecred is potentially quite useful for existing companies

fun fact on point 3, I was actually working at Google Brain at the point in time when I SourceCred coalesced in my mind. i quickly decided to leave Google to start SourceCred as an independent open-source project, but I mentioned why I was leaving to a very senior Google exec as I was on the way out, and he was really excited about SourceCred for that exact reason, and offered me resources + a team to build it internally so that it could be integrated with Google’s perf process. so there was corporate interest in that direction from the beginning

However, the goal for SourceCred is emphatically not to create a startup with solid revenue growth and to get myself and a bunch of VCs rich. the goal is to create the seed crystal that will recreate our dying economic paradigm and enable a new economy that’s more open, more inclusive, more sustainable, and more healthy. and IMO that’s fundamentally about open source

open source is such an economically radical movement – if you think about it actually realizes the marxist vision of giving the means of production to the people, not by shifting ownership from capitalists to “workers” (this just created corrupt statist super-capitalists), but by transcending the concept of ownership

i want a world in which everything has become open source: culture, art, technology, medicine, and yes, software.

and to me this feels structurally inevitable–its just economically rational, in that open-source technologies are more productive and more valuable than closed-source technologies. i saw this up close at Google Brain because I was one of the authors of TensorFlow, which is the technological crown jewel of the division of Google most important to their future growth prospects, and they chose to pre-emptively open source it, and that was a rational choice

they recongized that the structure of the technology ecosystem is such that the winning technology would be open source, and if they open sourced TensorFlow they had a shot at controlling the winning technology, and if they didn’t it would inevitably become a forgotten also-ran, because that exact thing happened with MapReduce vs Hadoop

you can also look at the fact that the internet’s backend runs on linux rather than microsoft as an example of open-source technology proving enormously more productive and valuable than the closed source alternatives

now, as we all know, open source doesn’t make any money

so in that sense deploying SourceCred on open source may seem like a dead end

insofar as SourceCred creates a cryptoeconomic system for open source projects. for the tokens to be relevant, there needs to be external demand for the tokens

within crypto, demand is mostly provided by speculators

this might seem concerning (“crypto has no real value, its just a casino for bored capitalists”) but this pattern is actually a structural feature of how technoeconomic paradigm shifts unfold: the kernel of a new paradigm cannot compete in the environment of the existing paradigm, because all of the competitors in the environment are co-optimized with the paradigm itself. however, in the closing days of the technoeconomic paradigm, there is enormous excess financial capital hunting for deployment, because the paradigm is exhausted so there’s no where for money to go. so the money streams into potential new paradigms, and provides the initial seed energy needed for the technology to get to the point where it can unlock value on its own.

read carlota perez “technological revolutions and financial capital” if you want to understand this dynamic.

anyway, it might seem like deploying SC on open-source is an economic dead end because open source doesn’t make any money, so who pays?

for this i have a short, medium, medium long, and long term answer

the short answer is that sourcecred will be generating a great deal of buzz in crypto at the time that it begins to spill over into open source

and crypto speculators who are accustomed to buying correlation with shitcoins and sketchy IPOs will start to get the opportunity to buy correlation with real open source projects and communities that are demonstrably creating real value in the world

so in the short term, there’s speculative demand because the speculators get a new (and much higher-quality) asset than what they have access to right now

in the medium-short term, SourceCred builds channels for existing (non-financial) value transactions in open-source

look at major open source projects and you often find FAANG etc paying ppl to work on them

React; facebook, TensorFlow; google, Linux: many parties, etc

FAANG is not spending this money altruistically, for technology that is complimentary rather than essential to their revenue capture mechanism, they reap a lot of economic benefits from open source

however, they spend this money in really inefficient ways

they do it by hiring people full time to work on the project

lots of friction, lots of inefficiency around hiring, tends to have high barriers to other contributors because the project is being built inside a centralized corporate information network

in comparison, sourcecred will offer a model of developing open-source software that is enormously more capital efficient

as an example, the total $ investment that’s gone into SourceCred is laughably small compared to how much we’ve produced so far

one of the results will be that much smaller corporate entities will start putting money into open source, via Grain

Here’s a specific example. I was the lead of TensorBoard, which is a prominent public facing part of TensorFlow and tended to get a lot of feature requests

My incentives were mostly to ignore those feature requests, and prioritize whatever I’d written on my OKRs

however, one day a feature request came in and i dropped everything to ship it inside a week. why? because the feature request came from a super high profile team within Google Brain that I knew was being closely watched by my management chain. By prioritizing and solving their issue, I got a lot of Cred within my organization. And this team got their blocking feature done inside a week.

The value to Google of that interaction alone was probably in the $MMs, since this team was blocked, and was a high profile partnership with important external people, that kind of thing

Now, imagine that the same team were a super flush VC backed startup (likely some of the external feature requests that I ignored had this character). Suppose that they wanted to spend $100k to get their critical feature added to TensorBoard, and then reliably supported long term.

How can they do this? The population of people who can make in-depth changes to TensorFlow is basically only Google employees, and Google doesn’t rent out their services. So there isn’t a liquid market for open-source work, not because there isn’t demand, and not because there isn’t supply, but because we don’t have the right value transfer mechanisms

People try to solve this with bounties but they are obviously inadequate because they don’t incentivize the creation of long-term value, and they are actually parasitic rather than sustainable for any open-source projects they’re used on

SourceCred, on the other hand, offers a sustainable way to mediate value flows in open-source that encourages long-term maintenance of the project, and flows upstream to support the whole ecosystem

I want to make this image into a meme where on the top it says “OS Sustainability” and on the bottom it first says “Bounties” and next says “SourceCred”. just need to find a gif editor. https://imgur.com/lgwvyqF(edited)

longer term after that, we organize mass relicensing of the open-source commons so that all open-source usage is free, but closed-source usage needs to pay a Harberger tax to their dependencies (or open up)

in the very long term, cred and grain will just be a natural economic medium for flowing value, so we’ll no longer need to ask where the money comes in

~end rant~

[AL0YSI0US]

@decentralion

“the goal is to create the seed crystal that will recreate our dying economic paradigm and enable a new economy that’s more open, more inclusive, more sustainable, and more healthy.”

That’s mighty fine looking goal my friend.

If only I could there’s a hiding in there.

[echojuliet6a]

Does SourceCred actually offer a more efficient alternative to HR or a more effective or altruistic distribution of resources than a traditional company structure? The evidence would seem to me to suggest that it does not.