292 karma · joined November 12, 2022
Change is also a constant, and there are plenty of good companies give bad, and a few in reverse. So due diligence is an ongoing process.
That dramatically raises the price of SaaS, so I use it much less than mainstream industry practice, but much more than zero.
CS has deep theory. IT doesn't.
It feels like we'd be suing ourselves. Wells Fargo defrauded us and if we win, we'll get money back from an entity we own
The might be a little bit of a shuffle with money going from people who bought one year to those who bought another, but for buy and hold, that's pretty moot. It seems like the lawyers are making money at my own expense?
What an I missing?
I use S3, EC2, Route 53, and RDS. That's more or less it. I advise others to do likewise. Not the same set, mind you, but the same mindset.
In truth, though, Amazon should have bought gitlab instead.
A good approach is to plan on all things, from plateau to linear growth to polynomial to exponential to singularity, as within the realm of possibility.
Or to have the humility to acknowledge that we may be wrong.
My lesson from the dot com boom is that's it's very hard to pick the winners. For every Amazon, there were a thousand pets.com. Investing in DEC, Sun, Yahoo, and friends would not have paid off.
That's a broader lesson too. I have an okay -- far from perfect but decent -- track record knowing which national economies will grow, but investing in index funds for those economies rarely yields good returns. I have a tiny portion of my portfolio in various developing markets, and the GDP can rise tenfold with no investment gains.
Many very real revolutions have bubbles. Bubbles are the period when people over-invest.
Software done right from 30+ years ago is still in broad use. There is huge value in that
Doing things right involves prototypes, iterations, and experiments, so high velocity is good. However, if a prototype ships, you often get on trouble.
I am at close to zero rush within group (creepiness, leading on, sexual harassment, etc.). Same type of casual physical contact across gender groups, and it's a trip to HR or a cancellation.
A lot of language takes on different making across cultures, and there are literally textbooks about how white casual language is perceived as disrespectful in many African American cultures. The reverse is casual African American behavior incorrectly perceived as aggressive in white culture.
I like looking at this in international studies since there is not the baseline baggage of racism or sexism, and Meyer is a good study.
- Minority affinity groups pull people from majority groups and decrease integration.
- Anti-discrimination/sexism/etc. movements often add social barriers to interactions (e.g. things I do within my identity group would be misperceived if done across)
- Affirmative action makes minorities feel like they don't deserve to be there (and often leads to resentment and other consequences)
Progress in the past few decades has been limited, so it seems like we're taking the wrong approach, but I'm don't have a better approach to propose.
Green fields, blue sky, what should we be doing to resolve the historical issues we have around sex, race, socioeconomic status, etc.?
I think looking to countries which made better progress might be helpful....
Developer: I am, in all honesty, glad for the experimentation. However, you should understand that critical issues include the ability to remix. I cannot use your code in my project. Ergo, it is not open, free, libre, or similar.
You're better off underselling and overdelivering. The term should also convey what this does. I like "source-auditable" or similar. "Source-first" says nothing.
You should also think about longevity. How do I know the license won't change next week?
Then my logic kicked in and I came to the same conclusion you did -- it's not open source, but source-available means people can audit that it's honest. That's a huge win over closed-source, especially for something privacy-focused.
I think the gut feeling comes from a long history of projects using source-available licenses in shady ways. The brain pattern-matches.
That would be good to address upfront. Simply stating "not open source, but source-auditable" might help address this.
Another good thing would be to post the price upfront. Right now, it requires an email to get to which is an antipattern.
These issues are cosmetic (but important).
This looks like a really good project.
However, state tests are annual. You don't want to measure the same thing. If you don't measure the same thing, you can't compare.
There are technically non-adaptive test designs which might work, but it's not what we're using, and they're a lot more complex than adaptive.
Being answerable to funding sources, tuition, etc. is a good thing. Having money to soon off from is not.
It's almost all bloat and overhead made possible by being over capitalized.
I'm not claiming that applies to Nvidia either way specifically, but it's a good sign in the generic:
1. They're efficient.
2. They've avoided the curse of overhiring (hi, Google!)
3. They can stay afloat even if business drops a lot.
All caveats apply. 20≠100, employees might include staff and grad students, etc.
Still, I would never advise anyone to donate to any institution like this.
Google maintains a monopoly and their brand. Profits are entire the of percent lower by the few power users who use all blockers. Big-O profits remain similar.
That contrasts with real competition, brand loss, or monopoly loss. That can lead to astronomically lower profits.
The short story is that most states assessments are designed almost as binary measures, to see if students are above or below a cutoff threshold set by Common Core State Standards. They're designed to measure schools, and in particular, to flag failing schools where kids aren't meeting standards.
They're very good at that, actually.
However, that's almost meaningless as a measure for kids well above or below standards, or unaligned to standards.
"Growth" is almost meaningless here. If I know one number is less than three and another less than four, I can't subtract them.
It's more mathematically fancy, but that's the jist. It gets even worse since measures constructs are highly multidimensional and different dimensions are measured each year. It's like subtracting apples from oranges.
It also encourages the wrong behavior. For kids behind, I'll get the best "growth" by discussing on grade level material and leaving gaps for what kids failed to learn before. I'll also do well to ignore my students who are ahead. Indeed, students who did week last year will inevitably hurt my "growth."
As a footnote, I would not call this a strong claim. Talk to a psychometrician and you'll see it's common knowledge.
If adaptive tests move beyond those few states, the problem goes away.
Growth measures do exist, but they require adaptive tests.
HP is at about a fifth of it's peak market cap, largely as a result of a similar play.
I'd short Google, but there's an old saying that markets can be irrational far longer than an individual wallet can sustain a short....
No one is suggesting Google should encourage anything. Openly user-hostile actions devalue brand and trust. That's doubly-true for power users who are thought leaders.
If Walmart has store security shoot suspected shoplifters and unionizers with a 9mm, discontinuing that policy would not be encouraging unions or shoplifting.
If you're a startup, don't do this. Maximizing the growth exponent does a lot more than tweaks on per-user revenue. Intentionally annoying users is bad business.
When companies do this, it's also good to look at PE ratios, since it suggests they don't anticipate growth.
With the exception of a few like OpenAI, I can't imagine much success over open source models.
If it can genuinely predict stocks, start a hedge fund.
Lawyers are a little more expensive than handling a click, especially at scale.
It's dumb, but that's the legal system.
That might be a good reason not to have a legal entity in the US but a jurisdiction with more sensible laws.