Can confirm this anecdotally. When I use my bike light in blinking mode, drivers are more careful around me every time. When I use the non-blinking mode, there have been times when drivers have clearly ignored my presence and I’ve had to wave my hand in front of the light to simulate blinking to get their attention.
Taste is indeed subjective (see what I’ll do next ;)), but even with a panel of 32 experts, this list is limited to a handful of OECD countries in their spotlight, with one exception.
I’ve tried a good number of roasters on this list, and only a few come within batting distance of Father Coffee (not to be confused with Fathers) in Johannesburg and Rosetta Roastery in Cape Town. Those two roasters, the former more experimental and the latter more traditional, are doing wonders and punching well above their weight in the South African market. Both could easily be in the top 25, and Father is surely top 5.
Seriously, if you care about coffee and find yourself in Johannesburg, go to Father and grab as many beans as you can carry.
Note for anyone in the UK or Europe: summer lilac (a type of butterfly bush) is highly invasive and spreads easily. In the UK consider planting native alternatives such as gorse which flower for most of the year. When gorse doesn’t flower, lavender will. For butterflies consider cow parsley.
I learned this when I had strawberries from the Bon Marche in Antibes. I was there in season, and the strawberries were small and “ugly” by supermarket standards, but my word did they taste like nothing else on earth. When you know, it’s like an awakening.
> None of the in-year deficit cited in the section 114 notice relates to the settlement of new equal pay claims. And none of the £149 million of cuts to services relate to any deficit created by equal pay settlements.
Instead, the council’s spiralling deficit, estimated at more than £300 million, appears to relate to the disastrous launch of the Oracle IT system that went live in April 2022. Originally budgeted at £40 million, the latest report by the head of financial planning to the cabinet on February 27 2024 showed that the Oracle finance and human resources system has now run to an astonishing £131 million.
<end>
In short Birmingham Council claimed that equal pay settlements are the reason for the bankruptcy, but the investigation shows that an Oracle IT implementation which has overrun its budget is to blame. The nefariousness is not just mismanaging the project, but casting blame on a scheme that’s supposed to correct more past wrongs. It’s turtles all the way down.
Try downtown Johannesburg, turning cars are supposed to wait for pedestrians when it’s a green pedestrian light, but the minibus taxis will intentionally drive through the crossings forcing gaps between pedestrians. You almost have to hold your nerve because if you don’t use the green, you’ll never be able to cross.
And of course being South Africa you also have to watch out for the “stopped” cars, which are never quite stopped and actually edge forward while the driver is checking their phone and not watching if they’ll bump into a pedestrian crossing. When I used to run there I had to tap the bonnet/hood of cars multiple times to catch the attention of the driver because they were edging forward over the crossing while checking their phone.
In London we have those but cars frequently stop over them either because the drivers are ignorant of the rules, they don’t care or they were too ambitious about making a green and ended up being forced to stop at the last minute. So as a cyclist I frequently have to push up to the pedestrian crossing for my own safety from cars behind.
I was in Montreal recently and the metro is one of the nicest I’ve used. It was on time, reliable and well designed. The stations are all architecturally interesting, and some of them even “beautiful”. The trains are very comfortable and unlike most metro’s, the lighting is soft and doesn’t overwhelm your senses when taking at night. Also for anyone interested in trains, the rubber tires on an interesting touch, albeit limiting if they wanted to expand the metro above ground.
> Fundamentally, it seems to me like there just aren't as many lung cancer screening scans out there for non-white patients as there are for white patients.
Just to qualify, you mean for the USA alone? It seems to me that part of the challenge is recognizing that the research needs to take place beyond just Western countries, or acknowledging it where such research is already occurring. Understandably many people would not be so comfortable with Google accessing patient data from around the world, so the next challenge is how diverse and global data can be protected so that important medical research can take place without any compromise of privacy.
The challenge is hard but surely not impossible, as this was the approach taken by the AstraZeneca-Oxford (and others) which conducted part of its covid vaccine trials in South Africa to test efficacy on non-white populations.
I’ve always thought that vaccinating elderly and first responders was less about stopping spread and more about reducing fatalities among the most vulnerable. After all it is the least vulnerable who are most likely to be the main spreaders, and until a reasonable portion of <65 year olds are vaccinated, the spread is likely to continue and probably even flatten before that point.
However since most countries in the world are in the middle of a 2nd/3rd wave, it makes absolute sense to vaccinate the vulnerable first rather than try race against the curve with a vaccination programme. Especially when the vaccination lag is longer than the spread.
> Unfortunately this is a wildly speculative argument, which is unlikely to find favour with regulators in this case.
I admit it is speculative but the argument encapsulates the abuse of the social network as I see it, which is why I'm running with it. It remains to be seen what angle the regulators will present when proving abuse of monopoly power.
However I cannot separate what Facebook is able to do on its social network without consequence from the idea that Facebook is an abusive monopoly.
> If I improve FB's/Google's ad models, do I increase the cost of all of their users? That seems ludicrous to me, but maybe that seems fine to you.
Consider the "user price of improvement". Suppose that an initial model m1 only needs to regress on a users age, gender and location to guarantee a clickthrough rate of x%. As the social network grows and there are more users, a larger variety of adverts and longer engagements, all else fixed the clickthrough rate will go down if the users start seeing more irrelevant adverts (more ad bidders but same relatively static input features) as well as repeated adverts (longer engagement times but same input features).
In an effort to increase clickthrough rates either beyond x% or at least to maintain x% in a growing network, the model needs more express power. So m1 is expanded to m2 now using accumulated likes. Fine likes are activity which is generated on FB so that's fair game. But there's a small cost in user privacy: suddenly adverts can be targeted based on your likes.
The network grows and clickthrough rates must still be improved. m2 is expanded to m3 by incorporating activity with friends, including mutual likes. Again this is generated on the social network, but there is a privacy cost: adverts are served based on public and perhaps private interactions.
The network grows and clickthrough rates must still be improved. m3 is expanded to m4 using browser trackers, so now your activity outside of Facebook is used to serve adverts: the privacy cost is increased and suddenly becomes real. FB serves adverts to you based on anything you do on the internet.
<side note: I recently ordered widget A using my laptop browser and the next day I saw an advert on IG for widget A from a competitor on my mobile device. I don't have an FB account and I'm not logged into IG on my laptop browser. I have never seen adverts for widget A on IG until this moment>
Back to my argument. Now at this point FB has a strong social network based on your activity and your friends' activity as well as external network effects based on browsing trackers, log-in with FB identities, as well as your friends' external browsing activity.
The transition from m1 to m4 has required using more and more of your personal data (whether public or private) in order to maintain a clickthrough rate of x%.
> This is a trade. You and I may think this is a poor trade, but it's one that many, many users make (and like) daily. In no sense is it a cost.
Each time the model is improved by adding additional features, those features aren't acquired for free. They're acquired by deliberately mining users' personal information. You might not think this is the case, because data is data and data is stored in bits and bytes, but it has representational value that in many parts of the world (most notably the EU) is protected.
In other words, the change from m1 to m4 in order to serve adverts has incurred a cost in terms of the private data a user must give up in order to receive adverts that are relevant to them (ie maintain a clickthrough rate of x% or higher).
So yes in this sense it is a cost.
Furthermore, in the 21st century, data is a clearly a resource, and has monetary value. It can be mined and gathered from users, which is then used to generate revenue from advertising. Users are not compensated for their data but instead are provided with a free service. If the alternative was to pay for the service, what would the monetary value be? That is the opportunity cost of providing data instead. So yes in that sense it is a cost.
Now let's consider consent. Facebook makes all users accept a privacy policy which constitutes as consent (the minimum requirements now may be higher but the tactic is the same). If the user disagrees they can't use the service. In most cases the user agrees because they want to use the service.
However as Facebook has a large social network and users' don't have many alternatives, the incentive for users to agree to the privacy policy becomes stronger.
If Facebook didn't acquire Instagram, then that last effect is weaker because users' have an alternative. If Facebook mined more data then Instagram, then Facebook is more "expensive" to use than Instagram and vice versa. Competition between the two networks would increase supply and reduce the cost of data required to use either network.
By acquiring Instagram and merging social network graphs (inputs to the model), Facebook effectively remove that dynamic, which gives them monopoly power to keep mining more and more data for marginal or perhaps zero benefit (an increase in costs).
Whether that dynamic would actually play out between users in a world where they are competing networks remains to be seen. Maybe it doesn't and for any network its a race to the bottom in terms of privacy regardless of the number of players.
> But consider the alternative. FB are blocked from buying IG, so IG need to set up a sales team, an adevertising method and epxand globally. None of these things are cheap or trivial. There does exist a world where IG competes on a level-playing field independently, but I would argue that at least 80% of the time, this hypothetical leads to IG either under-performing or flaming out.
Both of these outcomes are speculative, but I'm not convinced about the 80% figure. I'm not sure that IG needed FB alone to get to a billion users. Many of those users aren't necessarily people and are companies, organizations and other entities which use IG for engagement. It's enough that people spend more time on IG than FB the app or website to convince those entities to engage on IG. Not sure what the proportions are exactly but just giving an example.
Anyway, good points you've made and I've learned something so I'm willing to call it a day. I wrote this last essay simply because many of the points you made were simply too dismissive of what I said, without substantiation. Of course I'm not claiming to be right about how regulators will approach their argument, just stating a viewpoint on how FB is able to increase costs for the user, in terms of privacy, while being able to get away with it.
> Additionally, if (hypothetically) one chooses to make advertising-supported websites illegal, then that leads to FB (and Google) shutting down, which I don't see as a net positive for the world. (Certainly most of their users wouldn't think so).
Not sure why you brought up the prospects of advertising-supported websites being illegal. The purpose of regulation is to control abuse. GDPR in the EU does not making advertising illegal, but rather places limits on what information can be used. In the same way that road rules don't make driving illegal.
> The only google service that is really relevant and hard to replace is YouTube.
I would include Maps to this list. Outside of the US (and presumably China although I do not have experience here), Google Maps has no real competitors.
The luminance peak in the center is by design - that's clear in its parabolic gradient. Whether this is desirable or not depends on your use case.
On this point I don't see Turbo is being a do it all colormap. In the depth comparison images I still see Inferno and Viridis being superior and personally have no confusion between which spheres line up with which rings. Turbo in fact confuses me on this by introducing the false 2D color gradient.
Where I see Turbo being useful, as others have pointed out, is for heatmaps. I personally find Inferno's low end contrast being problematic and usually resort to Spectral for picking out details. If you can live with the fact that the lightest part of the chair in the test image is rendered darkly in Turbo (compare with Inferno), then Turbo is great for purely picking out detail (and taking brightness with a pinch of salt).
In the end, it depends on your usecase. Use a linear colormap when appropriate and vice versa.