ff317

Or Costco for really cheap food, even if you're not a member (I am one, FWIW). They don't require a membership card to go in through the out-door (because this is how you reach the customer service desk to sign up for a membership), and at busy times walking in through there is completely painless. The food court is in that same (post-checkout) area near the out-door, and also doesn't check membership. Most of the food is a decent value, but in particular you can get a huge all-beef 1/4lb hotdog and a fountain drink for $1.50, or their "chicken bake" for $3. I don't know why more non-members don't take advantage of it. They'd probably change things if too many people did, because they can't be making any money on those two items in particular.

It was and is about user traffic, not employees. WMF was at least implied to be one of the specific targets, because NSA had the Wikipedia logo on some slides about this program which were leaked by Snowden. Also, there was research showing the "chilling effect" - controversial articles saw less traffic in the wake of said leaks.

Right, this is the problem. The stock-market / macro-growth figures look good, but /personal/ finances and true cost-of-living look to be in trouble.

If the "solution" is a bunch of layoffs, it's just going to compound the problem. The solution is all these companies that aren't actually dying need to raise wages to match all the recent inflation.

I tend to think the problem is the "random digging for correlations" part.

Having tons of data is a Good Thing, so long as you can afford the marginal cost of gathering and managing all that data so that it's ready at hand when you need it later.

It's how you use the data that makes all the difference. If you're facing an issue you don't understand at all, don't go digging for random correlations in your mountain of data to find an explanation.

Think like a scientist: you need a valid hypothesis first! Once you have a hypothesis about what your issue might plausibly be, then you make a prediction: "If I'm right, I suspect our Foobar data will show very low values of Xyzzy around 3AM every weekday night". Only then do you go look at that specific data to confirm or refute the hypothesis. If you don't get a confirmation, you need to go back to hypothesizing and predicting before you look again. You can't prove causation by merely correlating data.

Yeah but that's also incredibly naive. Inevitably there will be some un-calculated-for shortcuts that make all those copies not as failure-independent as they seem. If nothing else, what are the odds some random Amazon engineer makes a software rollout or hardware upgrade mistake and takes out a whole lot of copies in the process?

In the 90s and early 2000s, it was not uncommon for me to hear some modem-like sounds from various devices with speakers (PC speakers, audio system speakers, car speakers). The sounds were basically interference from nearby (as in, mere feet away at most) cellphones of the era. I assume since then that cellphones have become less leaky and/or consumer electronics have become better at filtering out such interference, because I don't really notice this phenomenon anymore.

Honestly, the correct answer is that in the vast majority of cases, there was no good reason to ask any gender question in the first place. People should think about /why/ they're asking this. Even in the rare cases where there's a perceived need, it's likely a more-specific and functional question would be better. Dating sites are probably the biggest exception where I can imagine there's a legitimate need for putting a lot of thought into true gender+preferences questions.

Yeah... usually when I'm searching, I'm looking for something deep in the weeds, and what I get are really surface-level results. It's like I'm searching for some deep detail about how 3-phase grid voltage levels react to a severely imbalanced load per phase, and instead I get a bunch of junk about how to plug a lamp into a household socket without electrocuting yourself (totally made up illustrative example!). I truly think 1990s AltaVista was better at this kind of stuff than today's Google (but maybe not early Google).

A decent L2 charger should be installable at scale for ~$1K per unit IMHO. I'm basing this on the fact that a singular L2 home charger can be installed for about that. Figure the industrial variant costs a little more, but you get savings from the mass scale of deployment. They probably cost more now, but competition will bring it down as we scale.

Also, not everyone will even need L2 everywhere all the time, because many will be able to charge at home or use fast-chargers in emergencies. You don't have to be near-full at all times. You could deploy them at only 1/N spots, say something closer to 1/4 of all the spots, if even that (apartments might need 1/1, but streets and business parking lots/garages would need far less. You don't need them in any short-term street parking areas, as L2 is mostly-useless unless the car is sitting in place for hours).

You also don't necessarily need to have the raw power to run them all simultaneously: you can have local groups powershare (e.g. deploy 8 chargers with a feed-in that supports 4, and the chargers can coordinate to drop their charge rate as more people plug in).

If those wild assumptions are true ($1k, 1/4 of parking spots), LA's bill drops to $1.5B, which seems much more reasonable. The capacity will build up organically over time as EV adoption grows, starting with corporate and apartment parking lots.

I think in this hypothetical all-EV future, there would be other compensating changes to the city as well. Like, all gas stations would go poof, and their tanks, and the fuel delivery trucks, and most of the refineries, and all of the associated impacts on peoples' health from both the fueling and the car exhausts, etc. There's a lot of potential upside to offset any reasonable electrification costs.

Disengagements in Tesla data are going to commonly be for much less serious things than potential accidents (merely inconveniencing others, or embarrassing the driver in some way, or even a safe move that just didn't /feel/ right to the driver). They've published actual accident statistics for Autopilot, and those show that it has a lower accident rate than manual driving even on the same fleet of Teslas (which in turn have a lower accident rate than the rest of the US even when manually driven).

Driving is inherently very dangerous. Traffic accidents are a leading cause of death in the US. You're not really chasing perfection to win this game. It's not all that hard to beat humans on average, because the average human is pretty terrible. It's a statistical certainty that some people will die at the hands of Autopilot even when it's in some final non-beta state, but it will probably be less people than would otherwise die to the same miles driven manually.

The hard thing for Autopilot-like systems is perceiving the world accurately. "The world" includes both the physical reality of roads+cars it senses, as well as things like traffic rules, corner cases (construction, debris, manual police control of an intersection with a dead traffic light), and gauging the intent of other/human drivers. Humans are inherently better at that stuff. The software has to get better than it is today, but it will probably never fully match the best humans at this part.

However, there are two key ways the software inherently outdoes humans:

(1) It can have more sensory data than humans. Even Tesla (currently sans Radar + Lidar, just cameras) can see in every direction at the same time, all the time, with an overlapping array of cameras. No blindspots, no temporary blindness in one direction while you crane your neck to check in another, etc.

(2) It never gets tired, distracted, or drunk. It's never checking a facebook feed, or nodding off after a sleepless night, or too engaged in a conversation with a passenger to remember to look for that cross-traffic, etc. This is a huge advantage when it comes to accident rates.

I don't have it anymore, but for many years I kept an old Nokia CRT that I considered to be the pinnacle of the CRTs I had used. It was probably purchased in the late 1990s, something like (from googling) the Nokia 447xpro model. They used Trinitron screens, but Nokia also added really nice software for the on-screen control/setup stuff. I miss the old Nokia (their early GSM phones too!).

Ends-justifies-the-means reasoning has also lead to many of the innovative wonders we're all now relying on every day. While the customer test drivers aren't "trained", there was some caution in the approach.

Customers had to opt-in to request beta-testing access, they had to pass a Safety Score system that grades their driving safety (same as car insurance apps, roughly) for a long period (in some cases many months!), etc. After going through those hoops, when they finally get the software for it, they're required to consent again. IIRC the text there includes things like: You are in control at all times, must keep hands on the wheel and eyes on the road and intervene for safety, you are liable for anything that happens, this software can do the worst possible thing at the worst possible time, etc. They also monitor for your hands on the wheel (via torque sensing) and use an in-cabin camera to monitor whether you're watching the road or looking at a cellphone, etc. These measures can be defeated with effort by unsafe idiots, but that's no different than the risks such unsafe idiots present when driving any car.

With all of that in place, they've scaled up over a couple of years to 160K customer test pilots. Accidents happen, but there's no evidence the rate of them is anything to worry about. If anything, what evidence there is seems to point in the direction of FSDb test drivers being safer than average. However, they're supposedly removing the Safety Score part of the process Very Soon (likely in the next few weeks), but the rest of the warnings and mitigations should remain.

--- meta stuff:

There's a ton of money and a ton of ego pushing huge agendas in every direction when it comes to anything Elon-related, Tesla included, especially since the Twitter saga began and he started really going off the rails more. Almost anything you read on related topics, regardless of which "side" it's on, you have to question the motive to even begin to understand the underlying objective truth. I follow Tesla news a lot, and I'd say ~90% of all random internet news articles on these subjects (positive and negative) are mostly utter bullshit clickbait when they're not outright fraud, and they're designed to influence stock prices and/or buyer behaviors more than they provide useful information. When big money and big egos are in a war over something, objective truth on the Internet is a casualty.

If you ignore all that line noise and look at the objective reality of the engineering parts though: it's pretty amazing beta software with a lot of future potential, and the testing has gone pretty smoothly in terms of safety. It could be many years before you'd let it chaffeur some elderly person on a pharmacy run as a robotaxi, but IMHO it's still a better bet than most of its competitors in the long game of fully-generalized self driving on the human-centric road networks we have today.

As for Elon himself: clearly some of his behavior and viewpoints lately are both pretty objectively terrible. At least you can see it? How many executives from companies that built things we all relied on from the past few decades have really been any better? They've mostly been better at hiding it, while Elon puts it on full display. The world is what it is.

The main interesting charts they're showing are about disengagement rates, but this is a pretty sketchy comparison, both between manufacturers and over time with Tesla's FSD as well.

"Disengagement" is going to happen for different reasons for all of these different cases. There's the axis that runs from merely inconveniencing others (e.g. "I disengaged because I was holding up traffic by being embarrassingly slow through a turn at an intersection") to the ones where a serious accident was averted (e.g. "I hit the brakes moments before the car caused a head-on collision). There's the scenario differences that feed into thresholds for deciding to disengage: professional safety drivers on planned tasks that have been given a sort of "rules of disengagement", vs everyday normal humans getting groceries and intervening whenever they feel like it for whatever reason. There's the fact that many competitors are operating in tight geofenced, HD-mapped areas, while FSD is operating real-world all over the country.

I agree that Tesla's lack of transparency is troubling here, but this article seems to be trying to pressure them to increase it by taking a dishonest tack and stack all the unknowns against them in the worst possible way.

They have published basic accident statistics (in the past), and those have generally shown their automation to be a net win. In their own published stats from Q4 21 (latest available on https://www.tesla.com/VehicleSafetyReport ), the comparison they make is this:

* NHSTA data says the whole US averaged one accident per 484K miles.

* All Tesla cars on the road averaged one accident per 1.59M miles (so, ~3.5x better? This could also be caused by the profile of Tesla drivers rather than any real car safety difference - income, where they live, etc)

* Tesla cars with some form of Autopilot engaged averaged one accident per 4.31M miles driven (~2.7x better than non-Autopilot Tesla data, and overall ~10x better than the national average).

The counterpoint goes something like this (not that I necessarily buy it, but this is what I infer to be Tesla's reasoning):

1) We're only going to fully "solve" self-driving with ML techniques to train deployed NNs; it can't be done purely in human-written deterministic code because the task is too complex.

2) Those NNs are only going to come up to the necessary levels of quality with a ton of very-real-world test miles. Various forms of "artificial" in-house testing and simulation can help in some ways, but without the real-world data you won't get anywhere.

3) Deploying cars into the real world (to gather the above) without some kind of safety driver doesn't seem like a great path either. There's no backup driver to take over and intervene / unstick the car, and so far driverless taxi fleet efforts have been fairly narrowly geofenced for safety, which decreases the scope of scenarios they even get data on vs the whole real-world driving experience.

4) Therefore, the best plan to acquire the data to train the networks is to use a ton of customers as safety drivers and let them test it widely on the real routes they drive. This is tricky and dangerous, but if it's not too dangerous and the outcome saves many lives over the coming years, it was worth it.

Did you get rid of longjmp() too? longjmp() is far more perilous for similar reasons and purposes.

Defer is useful for code clarity in many situations IMHO, and it's easy enough for a project or code QA tool to ban its use for those that don't like it.

Another interesting way to split the difference: identify a need at $work that could really use a better open source solution that $work could use off-the-shelf. Go write that on your free nights/weekends as an independent open source project that's more-broadly useful to others outside of $work, and then when you're back on the clock, switch hats and make use of it.

With this hybrid approach, you're still advancing the cause at $work, but you're not doing it directly or exclusively for them, and the open source project will have visibility, scope, and utility beyond $work and your employment there. You never know, it might even be the start of future promising career opportunities.

There be dragons of course with how exempt employment and copyrights work in the US, though (as in, this might be a trivial case where if the side-project is successful, your employer can claim rights on it). You could come to some agreement with your employer ahead of time about the independence of such projects, though.

Of all the departments in the org structure of Wikimedia, the Technology and Product departments are the two with the highest individual headcounts. Combined Tech+Product staff is slightly over half of the entire org's headcount. Most of the employees in these departments are technical in nature (either engineers of some kind or direct supporting roles), and you can make a fair case that they're all necessary operationally for the sites (maybe only Tech is necessary on a day-to-day level, but you wouldn't make it much further into the future without Product's work as well).

I would argue that the other departments/teams you listed are also necessary to the long-term health of the Foundation and the sites as well. However, even if we hypothetically decided that your whole list of "non-essential" org departments/teams/functions were a complete waste of resources (and again, they're not, IMHO!), those still don't comprise all the "other half" of the org's non-tech/product headcount, as you're still missing other essential teams like Legal, Talent&Culture (HR), etc, which are also significant chunks (Legal alone is a little over 10% of staff and extremely important and impactful!).

As to your point about budget growth: it was a different site and a different Internet in 2006. The org is growing up and becoming more responsible. It's getting more resilient in the face of an increasingly hostile Internet, it's getting better at fighting technical and legal censorship battles around the world, and it's slowly replacing the externalities of burning out a handful of highly skilled and overworked staff with more robust and "normal" staffing where employees can have real work-life balance and vacation plans or departures don't cause undue risks, etc. We're also trying to reach out to and support the growth of underserved Wikipedian communities around the world (both in technical and non-technical senses) in ways that we couldn't even dream of in 2006. Increased costs and overhead are to be expected with all of this!

Even at today's Tech headcounts, if you compare us to any other major sites of this importance, scope, or even traffic level on the Internet, we're still operating on relatively-shoestring tech staffing levels, especially considering we don't outsource core infrastructure to 3rd party services or clouds, mostly due to our strong user privacy and legal stances.

The "Internet Hosting" line doesn't mean what some seem to think or imply it does. There's a ton of non-volunteer work that must happen to make any use of that hosting bill (software + infrastructure engineering teams, and obviously lots of other departments a functioning organization should have). There are valid debates to be had about how Wikipedia asks for and uses donations, but trying to imply that it only takes ~$3M/yr to "run" Wikipedia and the rest is useless bloat is hyperbolic fantasy and completely unhelpful.

Responsibility actually is just admitting fault, IMHO. It's accountability that's all about paying the cost for the fault.

This is exactly where those lofty Mission + Values statements come into play and why they're important when they're both written and lived well (as opposed to just some bland generic screed made from corporate buzzword bingo terms that everyone ignores).

The Mission is what you're trying to achieve (the profit motive, in some form), and the Values are the guardrails. They say explicitly what things you won't sacrifice (like human rights and lives) in service of the mission.

Having this stuff laid out clearly from the top down helps a lot with getting efforts to create and enforce those lower level processes taken seriously.

For me personally, I classify myself (whether it's valid or not!) as an experiential learner. I have a very hard time learning abstract things in purely abstract ways. I can read a book about a branch of mathematics and try intently to learn from it and totally fail. But give me one practical, real-world problem which demands some knowledge of that type of mathematics, and I might pick up the bulk of the same subject matter quickly, to the degree necessitated by the project. Whatever my conscious intentions may be, my brain throws a toddler tantrum about it and is like "Prove to me that I can use this for real or I'm not investing in it".

Rust is not a panacea. You can't just claim that this one emergency project somehow solves all the future bugs. Would Linux being rewritten in mostly-Rust help with some classes of memory bugs some of the time? Sure. Would there be a lot of other tradeoffs to consider, are there risks, would there still be plenty of kernel CVEs going forward? Yes to all of these.

It's the underlying prudishness at various layers (some of it is only historical, some of it lives in the present to varying degrees in various sub-populations) which makes the rebellious "Sex, Drugs, and Rock & Roll" thing a thing at all. If there was no prudishness in the first place, there would be no space for it to be popular to revolt against it.

As always, though, correlation != causation. It could well be the case that half of children (autistic or not) have some extra innate capacity for socialization, and that typical parenting choices don't provide those children as much social input as their brain wants at a certain stage of development, and hence they make up imaginary friends to interact with in order to satisfy their enhanced social needs. In other words, the imaginary friends could be an effect rather than a cause.

Yeah but Falcon Heavy is a proven platform. It can lift almost as much as SLS at a tiny fraction of the price. We could've done something like Artemis 1 with at most a few Falcon Heavy launches years ago for pennies on the dollar, right?

> There are other clues too: why do a small percentage of women have superior color resolution in the green spectrum (as if they had multiple green channels)

Apparently a small number of women have Tetrachromacy, where they have a 4th type of cone in their eye and can see a much broader range of colors, thanks to their genes. The gene isn't super-rare (I think one article claimed ~12% of women have the genetic condition for it), but the functional ability to perceive with all 4 types of cones is very rare. https://www.bbc.com/future/article/20140905-the-women-with-s... has a story about one confirmed case.

I'll take the third side of that bet: we've already so fundamentally altered "nature" to our whims, often in very destructive and short-sighted ways, that we're well past the point of trying to pretend like we're preserving any "natural" state of affairs ( https://xkcd.com/1338/ !). The best we can hope for at this point, barring a severe and abrupt population decline, is to continue engineering our way out of this mess as best we can, with the long-term interests of humans at the forefront.

Except that our social safety nets are practically non-existent. I bet if you made a middle-class comparison where you took a bunch of the typical EU social benefits, and equated the private costs middle class Americans pay to take care of them privately (healthcare, especially for a family, care of elder relatives, how much they'll need to save up to put their kids in college and retire even meagerly on their private income, etc)... I bet the Americans come out on the losing end of it on average.

Also one of several instances where the Breaking Bad writers put a lot of thought into amazing episode themes and titles: https://en.wikipedia.org/wiki/Ozymandias_(Breaking_Bad)

I still think it's important that scientists be factual about the uncertainty, though. The alternative whacko theories also have uncertainty, so if we hold ourselves to that standard, we can also rationally hold them to their even greater degrees of uncertainty. When we try to pretend that a current theory is 100% certain and factual just to win an argument with idiots, it invites obvious and IMHO reasonable criticism of all of science.

Science is not a set of facts caved in stone. Science is a process; a way of thinking critically and exploring the boundaries of reality in a way that makes a best effort at getting closer and closer to the boundaries of the knowable in a rational and reasonably-objective way. It often makes missteps and corrections along the way, and we can acknowledge that openly while also contrasting it to non-science junk that isn't even on the right track.