morelandjs

Here's the problem I see with it. Say you take multiple "top tier" theoreticians and task them with the same physical problem. Suppose they all make slightly different starting assumptions, and these assumptions lead to mutually incompatible conclusions. Which one is right? Who do we believe?

The whole point of testing theories is there can only be "one truth". If you let theories run rampant without testing you end up in logically inconsistent chaos.

All that said, theories should certainly be given time to grow when experimental validation is not immediately available. However, there's many in the community that take great liberty with that long leash to the detriment of physics progress.

I'll just add that I appreciate her cynicism because there is a staggering amount of bullshit in the high energy theory community. Embracing creativity is absolutely paramount, but people need to grow up and accept their own theory failures when they occur.

She's exactly right. Any science discovery that's worth its salt is falsifiable. If you can't validate your results there's a problem. It's the responsibility of every theorist to make testable predictions, and to communicate to experimentalists how they might be tested.

There's frankly not enough pressure within the theory community to disregard cagey theorists that actively avoid testing their predictions, or disregard the importance of testing in general.

Any time I see ML claims, I want to see rigorous validation. At the end of the day, whether or not your approach works isn’t debatable. It’s measurable. My prior is that everyone is full of shit, because most ML claims are.

Congrats! If you decide you want to monetize this, truebill sold to rocket mortgage for something like a billion dollars. I could see your features supplementing their current product nicely.

Little bit annoyed that your phrasing attempts to speak for the world on the "coolness of tech". I still find it cool. Neural networks that generate art? Quantum computing making meaningful gains? Apple M1 chips are absolutely bonkers... Try to get a little less grumpy and appreciate the amazing things that are happening around you.

You might add what your general career goals are to this thread. Could be someone with good experience interested in providing mentorship.

Known unknowns, and unknown unknowns. Still useful to quantify the known unknowns and compare significance of various events according to them.

Prefect workflow orchestrator: https://github.com/PrefectHQ/prefect

Suppose you could buy ads that are active every other week. I feel like after a few months you could compare the active and non-active weeks (perhaps excluding holidays) and have a pretty good sense what you are getting for the ad spend.

Sure you might have some opportunity cost, but this seems like a sensible strategy for a new company starting out.

I would like to get involved in activism to expose Russians to the horrors of what is happening in Ukraine, via campusing their internet presence with some of the war footage.

Does such a movement exist? What channels are still mutual accessible between the west and Russia?

Thurb off Google Russia. Looks to be one of their more popular websites.

It’s a tough field to become proficient in because you rarely get to validate your process. But I’m starting to get a better understanding of double ML, and it really does work in a provable way.

Causal inference has come a long way, e.g. double machine learning is pretty nifty.

I’ve wondered, instead of betting on “truth” why not bet on the outcome of a webscraping query, where the code is open source?

Example, I publish some code that scrapes the outcome of an upcoming sports game. It returns a True/False payload.

Bettors can examine my source code and see the exact time when it will be run. They can then place their bet on the code itself.

Sure the code could point to a bad actor and someone could pull the rug, but you could also take the consensus result from a few trusted sources, e.g. Google sports, ESPN, etc.

Definitely don’t need it, but it does introduce you to cool people.

This was the value I saw in my PhD. It has opened doors to jobs where the people I work with are brilliant and generally objective in their evaluation of competing ideas. It is an absolute treat to work in an environment like that. On many occasions, I worked with people who just knew more than me on a certain topic, and I was enthusiastic to take a back seat and learn from them.

1. Plan to continue investing in index funds on a schedule 2. Watching myself for lifestyle creep. Would like enough reserves on hand that I’m not a cornered animal if I lose my job. 3. Building up separate savings (brokerage) for a house. Will pounce a little early if the housing bubble pops. 4. Applying some healthy skepticism toward “too good to be true” investment schemes 5. Unplugging from the click bait cycle to preserve my sanity 6. Staying heavily invested in tech stocks. They seem to just scale better over long time horizons, and I’m in it for the long haul.

Maybe take a look at the orphan drug act for inspiration, https://www.fda.gov/patients/rare-diseases-fda

I don’t really understand the narrow definition of quantitative analysis referenced here and elsewhere. If you know your metrics are failing you, then you implicitly have a more advanced model of success. Define this more advanced mental model so it can be invalidated.

If your mental model is not defined narrowly enough to be invalidated, then how can you improve upon it? I think the lesson here was that the measure of success was defined poorly, not that quantitative analysis is flawed a practice.

I see a slippery slope in ML engineering, where to make something “production ready” engineers layer a relatively simple workflow in about 8 different technologies. I sometimes wonder if it’s preferable to just deal with occasional issues as they occur than to anticipate every deployment bug under the Sun.

I agree with the author’s take for the most part. That said, the best line managers I’ve had have done some IC work in the past of the same nature. The best PMs I’ve worked with vigorously attack the customer facing aspects of the problem and trust their technical reports completely on the implementation details, including pushback when the ICs see problems in the product scope.

You have to admit it’s amusing how public sentiment of block chain mirrors the price trajectory.

I personally do not trust research that does not have reasonably polished publicly available code behind it.

A strong result isn’t just the final number, it’s also the process how you arrived there.

I think most of us can agree that the bandaid fixes aren’t work for young Americans. The system needs a significant overhaul, both financially and culturally.

It’s not unscientific to change your mind when presented with new evidence, quite the opposite. It is, unfortunately, unpalatable as a politician to be uncertain, even when the facts demand it. I blame our culture of “leaders always know”.

It’s interesting to think about this from a Bayesian perspective. What is the posterior probability we expect the virus to have originated from a Wuhan lab, given the outbreak happened in Wuhan?

I’m open to web3 being a new and exciting phase of the Internet. I’m just wondering when it will come to me. You’d think if it were such a huge revolution it would start appearing organically in my every day web experience, but I don’t own any crypto, and to the best of my knowledge I have not yet organically stumbled into a new experience built on web3 innovations.

This looks incredible! Excited to try this!

Exactly. One of the fundamental tenets of science is reporting estimates with appropriate precision, e.g. significant digits. If the probabilistic interquartile range spans multiple months, why would I report the mean of the estimate in "hours of effort"?

What's frustrating is I've articulated these thoughts clearly to managers in the past. They nod their head and agree and then ask for the hourly estimate again and we all collectively share our disappointment when our estimate is inaccurate after the fact.

Nothing like running head first into a wall and expecting a different result.

My management suggestion? Frequently identify and advocate for simplifications, and modulate the degree of simplification based on progress to date. If people enjoy their work, they will work hard regardless of timeline hawking. So the variable factor is really the volume and complexity of work, not worker intensity or motivation.