Yup, saw that review. My takeaway is if one has knowledge and training level of Scott Alexander then this book has nothing new to offer. But since most folks don't so this maybe a interesting read.
More precisely than "plausible deniability", it is plausible EMOTIONAL deniability.
When you put enough bafflegab around it, you can almost ignore that you said something unpleasant. Because the part of our brains that processes for emotional content, doesn't process complex language very well. Hence the example with ten paragraphs of complexity to hide the pain of a major lay-off.
After I noticed this, I found that I did this. I reliably use complex language when I don't like what I'm saying. So much so that I could use readability checkers to find discomfort that I was not aware that I had!
And I'm not the only one to notice this. See https://www.youtube.com/watch?v=fpVtJNv4ZNM for George Carlin's famous skit on how the honesty of the phrase "shell shock" over time got softened over time to "post-traumatic stress disorder". A phrase that can be understood, but no longer felt.
Corporations have just developed their own special complex language for this. And you're right. It is emotionally dishonest. That's why they do it.
To be fair, the simple answer is not so simple within Google.
The issue is that Google achieves reliability by insisting on n+2/n+1. Globally your service is in at least 2 more data centers than is required for full load. In each region in at least 1 more data center than is required for full load.
If you're using the Google toolchain, all of the scalability and fallover problems are automatically handled by the layers that you're relying on. Which everyone expects you to use, because they are already integrated into the environment.
But if you go to use Postgres as a data storage layer, then you also need to take care of replication, failover, backup, and make sure that this is integrated with the automated systems that Google already has to detect when this is needed. Even after you've done that, people from outside of your team will need to be convinced that you've done that. Simply because you're doing things differently, you'll get extra scrutiny.
As a result, even if Postgres would have worked perfectly well, it is usually not the optimal answer for someone who is working within Google's environment. Don't think of it in terms of, "Does this do the job?" Think about it in terms of, "Can those in the broader organization easily certify that this does the job?" That certification is easier when you use standardized parts that are themselves already certified within the organization.
My guess is that your interviewer was aware of this. And was left with, "What about that question that I didn't think to ask you about?"
If you're interviewing at Google, the expected answer to the interview question can't be to use Google's internal tools. "Use Postgres" is the standardized, understandable answer for anyone outside Google who needs to solve Postgres-shaped and Postgres-sized problems.
No, that can't be the expected answer. And indeed, use Postgres was an accepted answer.
But when the interviewer keeps pushing back, that's an invitation for the candidate to ask why, and ask what about the environment might make that not a good fit within Google. Doing that gives a stronger hire signal.
FWIW, that was the second time I interviewed at Google. The first time, which resulted in strong yes across the board at L7, the first system design was to design Youtube Video Upload. The second was a more practical problem about replacing a high-volume logging component where correctness was critical but environment was space-constrained (i.e. no ability to run old + new in parallel).
Those were my favorite system design rounds ever, thanks to the problems being interesting and the interviewers also being very dynamic. It was also pre-Covid, so it was just awesome whiteboard design sessions.
I have been on a rough flight from Denver to Orange County, CA. Based on that flight, it does not shock me that a plane could be torn apart by turbulence. Instead I'm impressed that so few have been torn apart.
Our technology regularly does ridiculously insane things, and achieves jaw-dropping reliability. So much so, that everyone just expects it to work. And we no longer recognize how amazing it truly is.
The problem with general surveys is that you find out what people who don't buy your product think that they might like. Then they don't buy your improved product either.
A second problem is that we're unaware of what we responded to. That's one of the things that A/B testing reveals.
While some cases have been struck down, about 1/4 of people on the sex offender registry were minors at the time of the offense, 14 is the age at which it is most likely to happen, and this exact scenario accounts for a significant fraction of cases.
It is worthy of note that John Polidori's model for a vampire was, in fact, Lord Byron.
Lord Byron's death was a result of what the medical profession then thought that they knew about blood. Namely that blood-letting was a worthwhile medical treatment.
Building off of this point, consider the polynomial x^4 + 2x^2 + 2. Over the rationals Q, this is an irreducible polynomial. There is no way to distinguish the roots from each other. There is also no way to distinguish any pair of roots from any other pair.
But over the reals R, this polynomial is not irreducible. There we find that some pairs of roots have the same real value, and others don't. This leads to the idea of a "complex conjugate pair". And so some pairs of roots of the original polynomial are now different than other pairs.
That notion of a "complex conjugate pair of roots" is therefore not a purely algebraic concept. If you're trying to understand Galois theory, you have to forget about it. Because it will trip up your intuition and mislead you. But in other contexts that is a very meaningful and important idea.
And so we find that we don't just care about what concepts could be understood. We also care about what concepts we're currently choosing to ignore!
There is perfect agreement on the Gaussian integers.
The disagreement is on how much detail of the fine structure we care about. It is roughly analogous to asking whether we should care more about how an ellipse is like a circle, or how they are different. One person might care about the rigid definition and declare them to be different. Another notices that if you look at a circle at an angle, you get an ellipse. And then concludes that they are basically the same thing.
This seems like a silly thing to argue about. And it is.
However in different branches of mathematics, people care about different kinds of mathematical structure. And if you view the complex numbers through the lens of the kind of structure that you pay attention to, then ignore the parts that you aren't paying attention to, your notion of what is "basically the same as the complex numbers" changes. Just like how one of the two people previously viewed an ellipse as basically the same as a circle, because you get one from the other just by looking from an angle.
Note that each mathematician here can see the points that the other mathematicians are making. It is just that some points seem more important to you than others. And that importance is tied to what branch of mathematics you are studying.
Join an organization. For example every city has Toastmasters, most have several. Easy to find, and it is an excellent place to meet people. And you'll learn how to convert social anxiety into social adrenaline.
Do you have a faith? Actually go to church instead of just believing. Are you non-religious? Several strands of Buddhism can be followed as philosophy and practice without adopting any mystical beliefs. Vipassanā (also called Insight) and Zen are a couple of examples.
And how do you turn random people that you met into life-long friends? You can reduce the time investment by a lot. If you call someone on a spaced repetition schedule, you can make them internalize that the door is always open. Without requiring a large commitment on either side. And a spaced repetition schedule is easy to achieve - just think Fibonacci. I'll call you back in 3 days. Then 5. Then 8 (round down to a week). And so on. It feels like a lot of calls at the start. But it slows down fast. Over a lifetime, it is only around 20 calls.
Play around with it. If it was someone you met and hung out with on a cruise, maybe start at a week for that first call. Either way, you're reinforcing the idea that we like to talk, and the door is always open.
You can use a similar idea to keep people who move on from your workplace in your life. People always mean to stay in contact. Then don't. But with structured reinforcement, you can actually make it work.
A spaced repetition schedule for speedrunning the friend-making process?
If it works, it works, I guess. And in a thread about loneliness, that’s all that matters. But it seems a bit calculated rather than organic, which is what we think of as the platonic ideal of friendmaking.
Think of it as an intentional way to turn a spark of connection into long-lasting coals. It can't work without that initial desire on both sides to make it work.
As with all forms of cynicism, it has a grain of truth. And a much larger grain of truth than is comfortable.
reply