2,963 karma · joined March 23, 2018
I'm not sure how you square this with promotions (the main driver of performance reviews). No company who hired me five years ago could have kept me with a constant 3% increase, that is significantly less than I achieved through promotions, the rate would no longer be competitive.
I'm in a TS shop, 95% of our code is TS. I would assume that almost everyone we've hired knows at least one compiled language but that language might be C++, Go, or Java. The only way to write a CLI tool that anyone at the company can edit is to write it in TS and so, indeed, >90% of the CLI tools we write are in TS. Works fine.
We have a basic disagreement on facts here. My experience of the world is that very few songs have any substantial difference from the world as 1,000 other songs. With a small number of songs having the properties you describe. They are different in the way the patterns on the skin of an apple are different. Utterly unique, but not in a way that matters to consumers.
Diamonds are, perhaps, a good analogy here. Diamonds below a certain carat size are a commodity. They're ground up by the ton and sold for use in a variety of products at a cheap price (category 1 songs). Larger diamonds are not a commodity - a 45 carat diamond might be considered a national treasure (category 2 songs).
Corn is a commodity too, but I buy my sweet corn exclusively from one of a few vendors growing varietals that I enjoy more than what I can get at a super market (the varietal can be sold on the futures markets).
One definition of a commodity is:
> a mass-produced unspecialized product
My claim is that most lo-fi hip hop (and other genres) really looks like it fits this definition. The songs are produced in their thousands and have differences in the same way honeycrisp apples have slightly different colors and sizes, but ultimately are interchangeable.
Any further than this feels like it's nitpicking at technical details. I cannot tell the difference between lo-fi hop hop playlist A and lo-fi hip hop playlist B because the differences in the songs are not meaningful to me.
I think it's useful to bisect music into commodity/non-commidty categories.
Category 1 is things like Taylor swift's music which is not a commodity.
Category 2 is the music playing on lofi hip hop radio[0] is a commodity. 28,000 people are listening to the music on that channel as I type this and and approximately zero people know or care who made that music. The specific song basically doesn't matter so long as it fits the genre and meets a quality bar. They're analogous to handmade rugs - all slightly different but not in a way the consumer cares about.
So far as I can tell, most music that gets made is treated like it's in this second category: Generic Genre X song. Producing something that manages to break out of this commodity status is deeply unusual.
In anecdotal support of this, I often go to venues to listen to "bluegrass" (category 1) in much the same way I go out to buy an apple. I want the bluegrass to be good, just as I want good tasting apples, but the particular purveyor does not really matter unless it's someone special like the Punch Brothers (category 2).
I've watched 1,000 person orgs rot because the senior leader could not settle on a vision. He was "always on", smart, articulate, all those good adjectives. None of it mattered without the vision.
Maybe one day modules will hit the mainstream.
The data doesn't really say this. Two things to know about education:
- A large percentage of school outcomes are dictated by home life. Interviewing parents or just picking from the population of people who can afford $20k a year allows you to pick from a higher achieving cohort.
- Private schools don't have to provide the same level of services to children who require additional care. This is a significant cost saving. Most schools I've worked at had multiple staff dedicated to the care of just a handful of children.
This is the key - most of our writing on modern software comes from people building things with requirements that are unclear. There are pockets of the software world where detailed specs work wonders.
One of my professors in college had made a career writing these sorts of specs in a certain niche of robotics. It'd worked fabulously for him precisely because nothing was new or innovative, it just required exacting craftsmanship.
I think the term of art you're looking for is actually "command and control structure". This exact phenomenon appeared under communist regimes as well and is not unique to capitalism.
If you're interested in a real world example of an alternative structure read up on the corporation Mondragon - it's better but even large coops are not completely immune to this problem. Modern businesses appear to have an inherent need to focus large groups of people towards a singular vision and it's quite difficult to avoid the "toddler manager" effect.
A higher peak perhaps - but you can't be on MDMA 100% of the time, and taking MDMA effects that other time. Whether taking MDMA works out over the entirety of your life is a personal question - many find the high not worth it.
It is difficult for me to imagine Zig succeeding without a similar capability - which is to say I hope they do not follow through on this milestone.
I have seen engineers at companies 1/1000 the size of Google generate $4 million in differentiated costs savings (definitely would not have happened without them). It seems plausible to me that someone could generate 10x that at Google.
Next to that is a constant overhead of talks with leadership and supporting other engineering efforts throughout the company through things like doc review, chats, pairing, etc.
What he doesn't generally do is "productionize" features - code is generally exploratory in nature or fixes a serious bug (or class of bugs).
Add a language feature like the above to TS and you'd see adoption overnight. Pretty much everyone is happy to let their build system add additional correctness guarantees if its fast enough.
Sounds like the skills aren't very differentiated. IME it's quite rare for a business to miss differentiation in business impact in an order of magnitude.
Going to your earlier question, the candidate in question has many skills but is not overly specialized. He commands that price because he can point at the right place in the stack and say "There's $10 million in savings here" when everyone else misses it or "Changing the architecture in this way lets us ship 6 months faster" and he is able to do this repeatedly.
At the high end of the white collar market you're paid to deliver impact and your pay is determined by the quantity of the impact and how differentiated it is (no big bonuses for the guy who presses the "print a million dollars" button).
It also exists in a much more banal state lower down the ladder. I have friends who landed in the low end of the developer market. They literally cannot do most of the work I do. They don't understand DB isolation levels, they don't have the skills to design systems around eventual consistency, they are not familiar with basic distributed system problems like clock drift, etc. I'm not even an infrastructure engineer, this is just stuff I and my coworkers need to get through our days to ship product code.
Could my friends do this stuff? If they applied themselves, sure! But they're not going to. It's also not enough to learn on the job. We occasionally hire someone who doesn't actually have the skillset and they usually flounder.
Your suggestion proves my point! With social contracts I don't need to pay this fee, but by getting rid of them I do; I'd be getting charged for other people's anti-social behavior!
Stepping back, social contracts are forms of trust at the societal level. Any economist or business person will tell you that being able trust others dramatically lowers transactional and operational costs. I want to be trusted so that I can take advantage of those lower costs.
Even at fairly mature companies, fully utilizing a top tier ML expert may require the equivalent of 6-10 SWEs.
Anyway, while no one knows what the future holds, right now it doesn't seem like many companies are looking to actually take on AI as a core capability, they're looking to leverage open source tools. I'm at a large company right now that's betting pretty heavily on AI and we're thinking of bringing on 1 maybe 2 ML specialists over the next year. My friends at other large tech firms report similar plans.
There are lot of nice customer service things that are only possible if people don't abuse them: solid warranty/return programs, high quality phone reps, etc. But you need people who adhere to the social contract, the kind of people who don't, say, buy a $3,000 TV for a single super bowl party and then return it.
I actually desperately want some way to tell many of these companies "I am a responsible human being who respects social contracts" - all solutions I can think of violate my privacy but we'll find out what's worse...
There's also probably substantial savings to be had by moving to a single payer healthcare system which I think we should expect in the coming decades - support creeps up every single year.
Hear me, nameless internet stranger, that you might learn from my mistakes: This is not true. Efficiency is only worth money if it increases profits - concretely this means your efficiency gain must result in the following:
1. Delivering features faster
2. Delivering features with meaningfully higher quality
3. Delivering the same features with lower headcount costs
(2) is hard to measure, so you can generally only sell on (1) and (3). During the 80s the business world learned the hard way you can throw away a lot of money on useless efficiency. If that interests you I recommend "The Goal" by Goldratt for a fictionalized account of those learnings.
* There's also a subtler point that corporate finances may mean that even if efficiency is perfectly captured, a 1% efficiency increase may only be worth 0.1% of your salary.
All of this applies equally to businesses that were sustainable (or on a path to sustainability). The amount of money a business can responsibly spend on loss leaders like a free tier is dependent on growth rate and cost of capital.
Now that cost of capital has gone up and growth has gone down, the amount you can reasonably spend on loss leaders is comparatively tiny.
This does not mean we should expect fundamental changes in most business models going forward. Most startups did not take the Vision Fund approach to their business model that so fundamentally warped a few markets (Uber, WeWork, etc.)
- Healthcare providers will only buy cancer medication whose prices are low enough to be covered by insurance.
- Insurance scours the market for the lowest cost providers and states it will only cover that cost, no higher.
- If the cost of drugs goes up, insurance does not immediately renegotiate prices. Healthcare providers stop buying the drugs while waiting for low cost capacity to re-appear so they don't take a loss on the care.
- Drug companies go out of business because they can't sell their inventory, driving up prices further while reducing supply.
We really ought to have a government funded drug manufacturer...
This can be difficult to wrap your head around in the US because we don't have social rules anywhere near as strict as Japan and so empathizing with the experience is difficult.
Expect a lot more of this over the next year. We've all gotten used to generous free plans due to the era of zero percent interest and high growth; that era is ending (at least temporarily). Almost every SaaS business that has not already re-done their pricing tiers since the interest rate hikes is in the process of doing so. Oh, and ad-supported plans are going to get way more aggressive on the ads.