Tacherus
9mo ago
I’m not sure how true this still is but I think at one point Starbucks stood out as a relatively easy place to get a job but offered excellent surgery-covering healthcare.
Tacherus
9mo ago
It’s always difficult for me to appreciate what others might find interesting in my daily tedium haha.
I can offer some tidbits though. I’m on a small team (<5 ICs plus manager) and our daily standups are scheduled for 15 minutes. They routinely go 1-2 hours. Every day. I blame a mixture of managers trying to catch up on what’s going on and some on the team just…..always having a lot to say.
Over my career I’ve landed on a kind of “uncertainty principle” for data where the more basic and essential to the business a piece of data is, the more unknowable.
Three examples:
I once worked at a company on a pricing competitiveness project, so I asked the following question: “here’s literally a customer id. What are we charging this person?” And I got 4 different answers. Each was labeled “production” and I was told each was correct.
Another role I once had involved my company’s network so here’s a question: “how much traffic is on the network, right now? Where is it going?”
We’re talking like 10-20% error in these measurements. It’s wild to me. I think this definitely plays a role in how short of its promise “data science” has fallen. It’s a marginal gain and for so many companies I’ve worked with there’s been much much lower hanging fruit.
Last example was an ISP trying to improve the efficiency of their technician house calls. Question: in how many of the visits last month did the tech resolve the problem as opposed to merely doing something that was pointless and telling the customer the problem was solved, and then the problem just went away? Who knows!
Tacherus
10mo ago
I think it’s a combination of our worst tendencies but also a fallback plan.
If we’ve systematically criticized and torn down anything that could be worth making even the smallest sacrifice for, if nothing is worth putting ahead of yourself, where else would you wind up?
Tacherus
10mo ago
I think the script component is also immensely valuable. I’m increasingly convinced there is no “authentic” interaction except perhaps among those closest to us.
In general we can either have a script or we can track everyone’s status to know how to act as you allude to above. Interestingly I think that’s one lens to use for the identity fixation phenomenon. It seems to me something like that is unavoidable if we decide to throw out scripts and be “authentic.”
I remember my first time abroad alone in college for the summer. I was groggy from the flight and found myself eating a home cooked meal at my host’s house with his wife and I realized I had no idea what to do or how to behave!
Fortunately I had a script I could use. What I thought was irrelevant—say thank you for the meal, eat a sizable amount to show appreciation, offer to help clean up, etc.
What a relief to not have to re-derive from our relative statuses what to do so as to not give offense.
Tacherus
1yr ago
I agree and my thinking has changed a lot lately as a result. The way it’s come up for me is having people near to me who aren’t succeeding/hitting milestones/etc. and realizing how….completely useless a “root cause analysis” is for them.
There’s a place for it, but in practical terms if someone asks me for help with a problem, or I want to help someone, the answer never seems to pop out of having juuuuust the right analysis.
Turns out you usually have to actually do something to change your position and, right or wrong, clinging to “millennials are so screwed” is a cop out for someone trying to live a life.
Tacherus
1yr ago
I can confirm this line of thinking having been raised Mormon.
I’d never even heard of “priesthood of the believer” until many years after leaving and so to hear Mormons tell it there’s
-
Catholics with their obviously political and non-Godly nicene creed and claim of papal connection to god
-
Mormons with their claim to modern day prophets and inspiration
-
the Protestants who didn’t like catholic rules so just made some stuff up
Tacherus
1yr ago
You allude to it several times in your post, and it coincides with my general view of Srinivasan.
He can be provocative and creative but I think he has a tendency to ask “technology” to do a lot of heavy lifting in ways that don’t make much sense to me.
A particularly striking example I’ll never forget is him describing (on a podcast somewhere I think) how a constitution should be a git repo and amendments are pull requests.
My response to that and many other things he says is “ok, and?”
He uses what I think of as like a tool theory or framing theory in contrast with mistake/conflict theory:
If we used the right (technological) tools or terms or framing the problem solves itself!
It’s a seductive way of thinking and can create very clever looking solutions, but I find it’s because the real problems are glossed over or not understood, which makes the solutions vacuous at best.
It sounds to me like this book is very much in that tradition of presenting a framing as though it solves something.
Maybe I’m reading too much into it though.
Google's Gemini AI has had some....issues: https://notthebee.com/article/my-dudes-googles-gemini-ai-is-woke-as-heck-and-people-have-the-receipts-to-prove-it
It's since been taken down: https://www.msn.com/en-us/news/technology/google-explains-gemini-s-embarrassing-ai-pictures-of-diverse-nazis/ar-BB1iMJXf
I think this is best explained as what happens when you think your politics is not politics, but some kind of ground truth.
What this most reminds me of is what would happen if you tried to fit a model to the formula: 1 if the number is prime unless it ends in a 3 or it has a 7 in it and if one of those is true but it's divisible by 13 then also make it 1. You aren't "cutting at the joints" and neither do most political views, as evidenced by the examples in the links above.
Few things can bring that into relief more effectively than trying to produce an algorithm which reflects your politics. I don't think politics on the right would work any better to be clear, but as far as I know they don't seem to see computer algorithms as the way to advance their views.
You won't be able to gradient descent your way to your value system.
More options
Context Copy link