Hey folks, thanks for *check notes* ⊠waiting two months for the supposedly weekly series of Questions to frame my week.
This weekâs I will pose three Gojek-adjacent questions. Why specifically Gojek? Well, though you can pose similar questions in the context of other large successful companies/private endeavors in Indonesia, Gojek represents one of the most successful in recent memories. Thereâs this incredible (relative) concentration of talent, capital, and resources, with a culture driven by impact, that solves real problem for millions of people in Southeast Asia ⊠such that itâs often kinda weird to see the founders of Gojek not solve the local civilizational inadequacy that exists around them (or at least, not make a public stab at them). Ă la J. Storrs Hall, you can explain this seeming inadequacy in at least two ways: a failure of imagination & a failure of nerve. Through this weekâs questions, I want to make an at least marginal improvements at solving the âimaginationâ part of the problem. So, without further ado, âŠ
Q1 - What explains the rituals forming behind gig-driver communities in Indonesia (and elsewhere)?
Humans are really good at pattern-matching (communities of them, even more so), and itâs been the story of our civilization that we create elaborate stories about how the world works that not only are ruthlessly optimized for fitness but also to satisfy our own curiosities, that doesnât always correspond to reality.
Thereâs this general observation that we see when we have a group of humans interact iteratively with some opaque, complex systems and try to benefit from its interactions with them. At the limit, we expect âinstrumental rationalityâ and âepistemic rationalityâ to converge with each other (due to Dutch Book arguments), but as weâre embedded, bounded agents, we face a very real trade-off between having good, well-specified, parsimonious, predictive models of reality vs having quick, easy-to-calculate, portable, reliably-winning, biased-for-survival models for action. Some ways this have been remarked upon:
- Kahnemanâs System 1 & System 2 modes of thought.
- James Scottâs Metis vs Episteme, Illegibility vs Legibility, Anarchist Communes vs High Modern Authoritarians.
- belum-selesai
- âThatâs the wrong question and it valorizes white institutions and white ways of knowing and being and structuring society in really problematic ways.â - legendary twet
One manâs modus ponens is anotherâs modus tollens; and the usual modern way of responding when seeing these âfolk knowledgeâ, âkearifan lokalâ is to take the modus tollens like so:
- If we adopt this messy system of knowledge thatâs evolved against its local niches for an extended period of time, then we will gain fitness in our local niches
- We in fact have seen that groups of humans who have adopted this system of knowledge have lost, reproductive-fitness-wise ⊠we havenât seen their peoples, or cultures, or ideas take hold in our modern society.
- So, through modus tollens (i.e. taking the contrapositive of 1), we shouldnât adopt this system of knowledge.
The post-rat response to this is illustrated through Chestertonâs Fence: itâs unwise to destroy a fence you see erected in the middle of the forest when you donât know what the reason behind it being erected, it might be the case that the fence was put there to keep dangerous animals out. In other words, itâs almost likely the case that when you have an evolved system built through patchworks and not through some totalizing first principles, itâs not going to have some legible, cleanly-separated set of goals that it pursues. Throughout its exposure to its environments, itâs gonna optimize for multiple problems at the same time (just like how biology is so complex). âGain Fitnessâ is in fact a multi-objective problem, and when you do away with some systems of knowledge thatâs been adaptive for an extended period of time (but not adaptive when seen through e.g. its value in the World Economy), you might risk trading off against some valuable things that you wonât know about until itâs too late.

To conclude a bit, we know that true knowledge should converge with the fact that itâs useful, but as humans we face trade-offs between truth and usefulness (all models are wrong but some are useful), and we sometimes underestimate how useful some evolved systems of knowledge are, to our own detriment. This is essentially a problem of âHow Do We Deal with Wrong but Useful Knowledge?â.
So, all that above is a long introduction to the question that I want to ask here. Gojek drivers face this management-by-algorithm thatâs opaque, and acts as an intermediary, a market maker if you will, between the drivers and the passengers. And I claim that the above dynamics also exists in Gojek drivers interaction with the platform.
But first, a segue on why Gojekâs interaction with its drivers the way it is: one tendency that you see among âplatformâ tech monopolies is that they try to commoditize their complement, where (from Gwern),
âcompanies seek to secure a chokepoint or quasi-monopoly in products composed of many necessary & sufficient layers by dominating one layer while fostering so much competition in another layer above or below its layer that no competing monopolist can emerge, prices are driven down to marginal costs elsewhere in the stack, total price drops & increases demand, and the majority of the consumer surplus of the final product can be diverted to the quasi-monopolist.â
Some illustrative examples (still from Gwern),
Gojek also acts like this, in that it tries to commodify both its passengers, and drivers. On passenger-side, though, demand is already quite commoditized: you can easily treat different orders from different passengers as practically fungible, that only differ in one important respect (origin and destination location).
This is arguably not true when you look at the situation driver-side. Gojek came in an environment where the dominant players are fragmented, local ojek pangkalan with its idiosyncrasies that are little monopolies of its own. This means that the existing stock of drivers canât be treated as fungible & interchangeable. Drivers have different equipment, with their own unique local knowledge, and with different preferences that might be very sticky (e.g. certain drivers might only want to receive orders from/to particular locations). This is a problem because this makes (at least some subset of) drivers as price-setters, and not price-takers, which takes away a portion of the consumer surplus from Gojek.
On local knowledge of drivers
It used to be that one of the biggest barriers to entry to becoming a driver in a particular locale is your knowledge of routes and roads and landmarks, so much so that there were (are still) examinations that test aspiring driversâ knowledge of their city. But due to Mooreâs law making compute easily available in everyoneâs pocket (smartphones), and Googleâs monopoly on search making them able to provide free map tile & routing service for everyone (Google Maps), there is practically no more moat by having a back-of-the-hand knowledge of your city.
The pressure to commoditize drivers is also strengthened further by the fact that ride-hailing monopolies have increasing returns to the scale of the networks over which they have command. A big part of passengersâ decision to use Gojek stems from the fact that Gojek has many drivers (who often use Gojek exclusively), and the converse of this is also true. Thus, insofar that drivers idiosyncrasies induce very lumpy, unpredictable, and uneven supply; and insofar that this feeds back negatively into would-be passengersâ decision to use the platform; itâs in the best interest of the network to enforce some kind of standardization over its drivers.
To achieve the commoditization of its drivers, Gojek employs hard rules and soft incentives, such as
- An obligation to wear the characteristic bright green jacket and helmet
- Mandatory use of driver apps, even for functions other than receiving orders, like navigation, viewing earnings.
- Heavy penalty against taking orders off-app, this includes enforcing anonymized contact and funneling all driver-passenger comms in-app
- Heavy penalty against refusing orders (unspecified times) in a row for a period of time
- Ban against using other competing ride-hailing apps
- Platform-set fares, drivers canât negotiate prices with customers; this makes service predictable for customers and positions drivers as price-takers
- Centralized payment system, that reduces direct financial interactions between drivers and passengers
- Bonus and incentive schemes to even out distribution of supply, like bonus thatâs tied with the number of trips, or bonus on certain periods of times for drivers in certain cities
- Some gamified system of internal status
- Some algorithmic matchmaking procedure that ostensibly optimizes for some relevant performance measure (number of trips, rating of drivers/passengers, total rupiah amount, trip fulfillment, bid/order acceptance, etc)
Most of these rules and incentives are not obligatory, and their enforcement is not perfect, but the sum effect of them to the pool of Gojek drivers looks eerily similar to what it wouldâve looked like if the drivers were to be organized in a command-and-control manner like youâd see in most other companies. This is essentially the thrust of the argument that âDrivers as Workersâ groups have: if it quacks like a duck then we should probably recognize it as a duck.
This means that the view that we have of ride-hailing platforms like Gojek that are shaped like âGojek is just a place where willing sellers of the service of motorcycle rides and wiling buyers of said service meet and transact with each otherâ are at best incomplete and at worst a plausible-deniability veil for Gojek lobbyists.
On the partner (mitra) - worker (pekerja) distinction
But I still think that the partner (mitra) - worker (pekerja) distinction is meaningful here, or at least that they should exist in a spectrum here instead of a binary. The key thing that most Gojek drivers trade stability for is fundamentally choice. Gojek drivers at the end of the day are not forced to follow certain schedules, or take orders from certain places; and insofar that they are forced to do something then we should regulate them less like a partner and more like a worker. Weâve also got to remind ourselves that Indonesia pre-Omnibus Law for Job Creation (Cipta Kerja) had a very rigid labor law, one of the most generous in the world for the amount of severance & redundancy pay it mandates to be given to workers. The fact that Gojek has scaled so quickly is also an evidence towards there being a huge latent supply of willing drivers that couldnât provide the service because there were no visible/contestable markets they could offer their service in.
This is also not to say that the commoditization itself is bad. In fact itâs a central theme of humanityâs progress over the past few centuries that we achieved them through the slow and gradual unraveling of barriers to trade amongst ourselves, enabling labor and capital and technology and information to flow to where theyâre needed the most. I want to continue to remind people that at the very least the first order effect of Gojek is good: there are millions of contrafactual transactions totaling trillions of rupiahs in economic value that wouldnât have happened if it were not for the existence of ride-hailing platforms like Gojek.
Again, we have so many things (at least in Indonesiaâs big cities) that we now take for granted. belum-selesai
What makes Gojek (and other ride-hailing platforms, again to be clear) unique relative to other âplatformâ monopolies here is that they employ opaque matchmaking algorithms heavily (unlike e.g. Tokopedia, or Amazon, which convey more information through reviews and prices). And unlike with explicit rules/incentives, itâs often unclear to the drivers why (interpretability moment) the algorithm nudges them to act out in a certain manner. And viewed in this light, itâs pretty clear that this story is very similar to the story of alienation that played out in the industrializing economies of the 19th century1.
Thereâs argument to be made that the transition should be made clearer to its drivers, and I can see how the argument for greater recognition for drivers as workers resonates emotionally here. In the long run weâre all dead,
With all that being said, whether you think this Great Commodification of Drivers2 is (net) Bad or Good, this phenomenon (of Gojek Drivers ) results in a few interesting things, that Iâm asking questions about,
- abc
(his is not to say itâs bad, it is in fact good to have a place where wiling; but here my argument is that the way that Gojek enforces willingness is )
- An ethnography of Gojek drivers, in different cities etc
- belum-selesai
On how we should deal with network-effect monopolies
Contents One of the reasons why we tolerate monopolies is that [abcd] belum-selesai
In this light, itâs amazing that some proposals to curb the bad effects of ride-hailing monopolies like Gojek often extraordinarily miss the point, and in consequence seek to
- Price cap and floor
- medallion stuff (like airport sticker thingy)
that ben kuhn article
talk a bit about how interoperability is such a good idea
like usb c in eu, but also worry that this will stifle innovation
add your twet here
- This means that unlike the usual markets that we often encounter,
- gvgvhg
This poses a problem because this makes supply very lumpy and unpredictable. A bigger problem
One pressure that Gojek also faces as a âplatformâ monopoly is that they rely heavily on network effectâwhich naturally tends to [natural monopoly] etc.
due to some realizations
- Gojek (& Grab) is not really a market
- But Gojek is commoditizing their complement?
- What does Gojek sell really?
- âexplainsâ is doing a lot of work here, and there are several interconnected questions behind the question above, which include
- dsaf
- Not all information in the market are conveyed through price,
- why does algorithmic management feel so oppressive? alienation
- Network-effect monopolies are not that free market as theyâd like to present themselves
- they are biasing themselves towards more networking in the platform - and as such, they heavily penalize behaviors that go against the lowering of marginal cost of (gojek faces this kind of problem where you need a critical mass of driver & customers to make the market self-sustaining)
- and it seems like itâs the revealed preference of a big chunk of the riders that they want [abdefg]
- mengapa gojek tidak interoperabel dengan grab, mengapa instagram threads tidak dengan twitter, etc
- Iâd like to do an ode to gojek sometimes, that even though with all the supposedly (and oftentimes justifiably) oppressive features of gojek, gojek truly enables a better life abcd; that thereâs a very heavy status quo bias, that makes everything becomes normal
- tidak ada diferensiasi yang jelas antara gojek dan grab, marketnya udah spent,
- gojek dan grab yang bersaing antar satu sama lain daripada driver individual yg bersaing satu sama lain?
Q2 - Why are there no Gojek-sponsored professorships/tenure positions in any Indonesian universities?
Like google with search monopoly, they were able to abcd But this is a bit unfair because Gojek is unprofitable at the moment But abcdefg
how google does research https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/38149.pdf
itâs not incentives, itâs you (that tal blog)
- This is framed more like a rhetorical question, in the sense that I think there should absolutely be a
- Also because Gojek represents a wealth of
- etnografi driver gojek grab
a more academic/mathematic/operations research treatment of their matchmaking algorithm, some sort of cross pollination would be useful??
Q2 - Was the âMerdeka Belajarâ suite of policies successful?
- I want to know the post-mortem
- Also norm of public works
- Also because Nadiem likes âhyperiteratingâ
Iâve been meaning to make The success of market
Footnotes
-
Theories about Alienation are essentially theories about encountering an alien, deterritorializing force that knows no bounds, no taboos, and profanes everything it touches.
(Ă la Nick Land - Capitalism as an alien entity reaching out across time seeding itself to humanityâs past to assemble itself in the future) (which is crazy because how intelligent something must be to anticipate the various ways humanity could evolve in the future) (the more boring explanation is that Capitalism is just like pipes, itâs an instrumentally useful way of allocating resources, just like pipes are a useful form factor to move liquid)
You can understand phenomena like Luddism, which I struggled to understand before, why would anyone be against increasing productivity (âàȰ_âąÌ). But this is easier to understand when you see this a first-contact story, of a group of people meeting an alien force that is the constant flattening into money (âexchange valueâ as Marx would term it) of everything by the market.
I have some sympathy towards this view of society, but we often have very romanticized view of the past, and deep bias for the status quo, such that itâs often hard to really grok how improved the life of the median person has been since our economy grew exponentially (i.e. how it seems normal to just expect total economic output to grow a fixed X amount of percent), and how we can still be better. â©
-
Marxists think this is very similar to the concept of reserve army of labor, whereby capitalists deliberately keep a surplus of workers in deteriorating conditions to keep labor costs low - which tbf again has that conflict vs mistake theory texture to it.
It might have some grain of truth to it? In that one of the key factors that enabled Japan to succeed in its textile industrialization was through keeping wages depressed whilst productivity went up, i.e. doing away with the prevailing piece-rate mechanism to compensate labor and instead replacing it with hourly wages. â©