3,607 karma · joined July 21, 2009
It’s much more effective to work learn about people, their predilections, biases and experiences. No one is a “checkbox” asshole. Much better to focus on specific behaviors, instead of writing off a person as a type.
When I see a company describe its “no assholes” rule, I see a company inclined toward calling people assholes.
That said, one of our folks suggested same so I think from a UX perspective, we could go either way. It doesn’t _feel_ like a form submission, one could argue.
Using self-identified rich to derive an income baseline for rich still means “rich” is self-identified, no? It still puts us in the 1-2% range.
I suggest that a measurable definition of rich should be those who are clearly above the vast majority of (say) incomes. Being richer than 80% or 90% of the population makes one closer to the top than the middle, so that seems like a better definition of “rich”. Agree that it is necessarily arbitrary, but I think such a definition is pretty conservative.
Regarding the 90%, it’s from this line:
> about 9 out of 10 respondents identified as being in one of the three parts of the middle class
But re-reading, yes you’re right that the lower bound is not self-identified. But the author’s using the $30K number (from the poverty line) puts us at around the 5th percentile: https://en.wikipedia.org/wiki/Household_income_in_the_United...
…which effectively puts the “middle” classes at 90% of the population.
Again, thanks for the rigor and let me know if I have misread.
The study appears to define "middle" as 90% of respondents, which is laughable. Everyone thinks of themselves as middle, no matter how far from the actual middle they are.
The obvious definition would be percentiles of income, or wealth, or other measurable advantage. This would put a lot of readers in an upper class, as which people really don’t like to think of themselves.
In my example, I was using iMessage and it captured my SMS in the same UI as IP chat. Then I changed to an Android phone, and my SMS would not arrive on that phone -- same #, same carrier. The issue was Apple's arrangement with the telco, and it required a support call to get that magic undone. Took days.
SMS and IP chat are different. The former is device-to-device, the later is user-to-user. The former is fire-and-forget, the latter has presence. Abstracting over these is a bad temptation.
That said, Project Fi actually gets it right. It make SMS and IP chat transport-agnostic, but at the network level, not papered over in the UI. It fulfills the promise.
Ride-sharing networks are O(n). They benefit from liquidity, but I derive only secondary benefit from being “connected” to others on the platform. It's not nothing, but it feels linear, not superlinear.
Ride-sharing apps are just apps, and the barrier for adopting a new one is low, for both sides of the market. Further, Uber’s political operations benefit the other apps equally.
Uber can defend its brand and its satisfied users. Size helps here. Think iTunes. But the “network” is replicable by others.
Big categories of software go underdeveloped, leaving an industry in the dark ages. It’s one reason why medical software is so bad – being on the wrong side of legal risk is too dangerous.
Good managers take crap so others don't have to.
Ideally, we get good enough to prevent the crap from happening in the first place. But it is most certainly the job. It's entirely understandable if you find it's not for you.
“He was diagnosed with flu” is passive voice, and appropriate, because the diagnoser is not important. We care more about the flu than the doctor.
“Mistakes were made” is evasive and serves to obscure, and should be avoided.
And in any case, there is no reason to believe that Yahoo’s or any tech company’s continued existence is normal. Yahoo was of a time, as was Digg, MySpace, etc.
It’s easy to say they “screwed up”, but success for any amount of time is the exceptional part. Regressing to the mean of “non-existent” or “not valuable” is…normal.
More importantly, it would make doctors (and regulators) more honest participants. If a doctor wants to recommend opiates, it would largely remove their economic incentives for doing so. We don’t think of doctors having much economic incentive for prescribing acetaminophen or ibuprofen, by comparison.
FDA approval is the first layer of prohibition. Very expensive, this keeps a lot of competitors out. Lots of market (read: pricing) power is conferred to the winners. Great incentive is provided for regulatory capture.
The second layer is patent protection. Again, competition becomes illegal, with outcomes similar to above.
Third is scheduling it as a prescription drug. The incentives of the drug company are now to influence doctors, in ways that may be more or less overt. Docs get a cut of the outsize profits, de facto.
The fourth layer is opiates being illegal in general. Competitors out.
Of course there are legitimate reasons for prohibition, even if I don’t agree with them. One can make an argument for protecting people from harming themselves.
But these arguments are naive and static. By “dynamically scoring” the cost of prohibition – following the incentives and the outcomes – we might choose differently.
The countdown clocks are generally accurate, but in certain situations they are surprisingly off.
I would love to see someone use this data to improve arrival predictions – a regression, say, based on features such as time of day (a proxy for crowding), weather, holidays, nearby events (concerts & sports), maintenance or signal problems...
I assume the clocks use something as simple a static numbers representing the time between adjacent stations or segments. They presumably don’t tolerate deviations from crowding or other unexpecteds.
For example, many categories of police activity (raids, traffic stops) will become harder to justify; the dollar cost of drugs going down by an order of magnitude; immigration enforcement.
I am sure that most of the personnel policies that the author alludes to are well-meaning and intuitive. What is out of scope in such rule-making is the empirical outcome. The cost of delay (213 days to hire) is apparently not a factor in the rules.
They lose a lot of good people in a presumed attempt to ensure they hire good people. Such an outcome is only recognized a posteriori.
That said, Java (and C#, which is where my experience lies) are fundamentally bureaucratic languages. Things like hierarchical namespaces and several layers of accessibility very much mirror their human-org counterparts. Explicitly implemented interfaces are a way of saying “I comply”.
Go, by contrast, is much flatter. Which, perhaps, is where organizations are heading. And interfaces are implicit, which is a way of just doing the job instead of declaring compliance.