There’s a debate on whether 10x software engineers exist.
They do: I’ve seen several of them.
And their existence freaks the hell out of me. 5 examples of 10x engineers and why you should be afraid when you see one:
1. The Move-Fast-And-Leave-Behind. A dev with a hacks mindset at a scaleup. They get shit done 10x faster than the engineers who that take this (literally) shit over when it needs to scale, try to reverse engineer it, but ultimately have to toss and rewrite the whole thing.
2. The That’s Trivial To Finish. Someone w many product-minded traits blog.pragmaticengineer.com/the-product-mi… amazing at prototyping and telling the non-technical manager they’ve done 90% of the work, and the other devs should have no problem finishing the last 10%. Which then takes 10x as long.
3. The Only Non Quitter. A company a terrible eng culture and just as bad codebase which oversells itself. Devs quit all the time and the new joiners struggle with everything. Save for TONQ who gets stuff done. Obviously the most tenured dev, and the only one lasting >2 years.
4. The Debugging Machine. A place with a codebase w no tests or documentation. New joiners tend to break everything and TDM needs to be called in to save the day. An engineer who has been around for years, though refuses to ever document/share any of their well-earned knowledge.
5. The Story Point Hoarder. A company where productivity == story points shipped. A tenured engineer who figured out how to make sure every second sprint they claim 5-10x as many story points as most other team members through cherry-picking work, optimising for these points.
So yes, 10x engineers do exist. They live in a mostly unhealthy engineering environments allowing for 10x behaviours.
If the above examples proved anything it’s how we should not ask: “how can we have more 10x devs?”, but answer “why are most our devs at 0.1x productivity?”
10x devs share the trait of being tenured at a company, and being perceived 10x as efficient as most new joiners.
Which begs the questions: 1. Why does an engineer need years of work at the company to get productive? 2. Is perception == reality?
Those are the 10x questions.
2 more archetypes: 6. The Reinvent The Wheel Dev. One of the first engineers at a startup who decides to reinvent the wheel. Writes a custom SPA framework, with layer, MVC abstraction. Then gets everything done 10x faster than new hires (who they label as “not smart enough”)
7. The Stupidly Hard Worker. Typically someone who is also #1 or #6 at some level. They work 12+ hour days, also through most weekends. Management loves them as they’re clearly devoted to the company, and ignores any complaints because this hard work & perceived 10x output.
Finally, my observation on what a highly productive engineer can look like (who I would not call 10x):
And yes, Sonos used to have a great software experience.
I got my first Sonos around 2019 or so I think - and the setup and tuning were very nice (positioning speakers in a room for best performance.) Worked well for me at least.
Major banks skipped due diligence on the deal when providing massive loans to the world's wealthiest person buying Twitter for $44B, assuming they would make a quick buck by selling on these loans.
But they cannot sell it on and make money on it?
The full story by WSJ:
It's hard to feel sorry for massive banks that don't make the quick buck they expected to do, because they loaned for an objectively terrible deal? (Twitter was sold for 2-3x the value of Snap, despite fewer users, similar rev)wsj.com/tech/elon-musk…
FWIW Snap today:
- Has ~2x as many users as we can assume X has (Snap: more than 800M MAU)
- Has ~2x as much annual revenue (about $5B)
- Is worth $15B
... meaning X would be valued no more than $15B today, most likely.
It's notable that coding assistants like Copilot, Tab9 and many others are available in most IDEs... save for XCode.
This means we have an unlikely "control group" to determine if these AI assistants make a major difference in coding: native iOS devs vs everyone else!
Assuming these coding assistants provide a meaningful and long-term productivity boosts: teams doing web and Android development using these tools (e.g. via Jetbtains or GH Copilot) *should* be meaningfully more productive vs iOS folks.
Interesting if we'll see major differences
The reason for this is how XCode seems to be deliberately hostile for extensions: and so the inline coding extensions that IDEs like Visual Studio and Jetbrains IDEs support (and that AI tools use) are not available for XCode.
Outside of coding and customer service, what are areas where GenAI / LLMs result in very clear productivity gains or business gains, without a deterioration in the experience for customers?
These are two areas I currently see as "yeah, GenAI actually works here, not just a fad"
Funnily enough, even when Sundar Pichai was asked about GenAI, he seemed to only list these two examples. Two weeks ago he said:
"There are pockets, be it coding, be it in customer service, et cetera, where we are seeing some of those [GenAI] use cases seeing traction"
The "et cetera" is what I'm interested in.
Coding is a fantastic fit for GenAI:
- Simple grammar (simpler than human language!)
- Huge amount of extremely high quality training data (code that compiles!)
- Hallucinations can be limited by compile/test
- Humans review output
Here is an EU regulation that surely massively accelerated online businesses:
The right to return any physical goods purchased online within 14 days.
Here’s why (my recent story with a faulty vacuum cleaner that will make me only buy stuff like this online, even from a shop:)
I needed a vaccuum cleaner while in Hungary. So I walked into a retailer shop and bought a cordless one.
The vacuum cleaner broke after 7 days (no charge.) Took it back to replace it… but was told that in-store purchases are not eligible for the 14-day return. Only online ones
So now the retailer is sending my brand new vacuum cleaner for repair. I have no appliance for 2-3 weeks while they attempt to repair what should have not been broken.
There is zero point as a consumer buying appliances in-person: thanks this 14-day policy not applying to them.
Delta (by regulation) needs to stand in for the losses and cover them for passengers.
The interesting parts of the lawsuit will be:
1. What contracts did Delta strike with CS/MS in case of them causing financial damage to their business, like now?
2. What does the judge say?
FWIW suing Microsoft seems to be pointless to me. Liability stands with CrowdStrike: they very clearly caused the damage. Just me, but I cannot see a judge come to any other conclusion.
Ship changes that run in the kernel at your own risk, as we will see.