No dog in this fight, all props to the Fly.io team for having the gumption to do what they are doing, I genuinely hope they are successful...
> It's still 99.99+% SLA
But this is simply not accurate. 99.99% uptime is < 52m 9.8s annually of downtime. They apparently blew well through that today. Looks like they essentially had the equivalent of 4 years of 99.99% uptime equivalent this evening.
Four nines is so unforgiving that it's almost the case that if people are required to be in the loop at any point during an incident, you will blow the fourth nine for the whole year in a single incident.
Again, I know it's hard. I would not want to be in the space. That fourth nine is really difficult to earn.
In the meanwhile, <hugops> to the Fly team as they work to resolve this (and hopefully get some rest).
99.99+% SLA typically means you get some billing credits for the downtime exceeding 99.99+ availability. So technically do get a "99.99+% SLA", but you don't get 99.99+% availability.
Other circles use "SLO" (where the O stands for objective).
You're describing conditions that occur in many asymmetric/guerilla wars. None of these are novel tactics whose acceptability must be evaluated from first principles now.
Further, none of these should come as surprises to Israeli commanders, who will have seen these tactics from Hamas in the past.
The bottom line is that any military can only control its own conduct as it represents its citizens in battle.
> You're describing conditions that occur in many asymmetric/guerilla wars. None of these are novel tactics whose acceptability must be evaluated from first principles now.
Some of those conditions are similar, some aren't. In most cases, the group doing guerilla warfare isn't actively trying to get their own citizens killed, or if you want to be generous, simply doesn't care if they get killed or not.
That said, you're partially right that these conditions have occurred before. That's why many military experts make comparisons to similar situations, like parts of the Iraq war or even closer, fighting against ISIS.
In most of these analyses I've seen, they claim that the IDF performs as well as the US army did in similar situations in terms of protection of civilians, civilian to combatant killed ratios, etc.
> Further, none of these should come as surprises to Israeli commanders, who will have seen these tactics from Hamas in the past.
I don't think anyone is surprised by how Hamas is acting, except much of the international community who simply refuses to accept how Hamas is acting.
> The bottom line is that any military can only control its own conduct as it represents its citizens in battle.
Yes, but if there are legitimate military goals to achieve - and there certainly were legitimate goals to achieve in the beginning of the war - then the military has to fight the battle its enemy is giving it. There simply isn't a way to fight Hamas without inflicting civilian casualties, because of the way it fights. You can choose not to fight it at all, but that wasn't really a choice that was available to Israel on October 7th. (Whether the war should've continued for so long is a different matter.)
It looks similar to a cable/fiber rollout where they have to onboard each region individually. I know they are currently doing this in the Atlanta metro.
Like cable/fiber, once they have good models of the business and what it costs to roll out, they have the freedom to accelerate and do regions in parallel. If the business works, I would expect them to scale the pace of rollout.
> the vast amount of training data they are collecting
They keep pushing this point. And they do appear to be collecting an absolute firehose of data from the millions of vehicles they have on the road. By comparison, Waymo collects a lot less data from many fewer vehicles.
Which leads to some tough questions about Tesla's tech. If they have (conservatively) 10x the training data that Waymo has, why can't their product perform as well as Waymo? Do they need 100x? 1,000x? 10,000x?
Assuming they were at parity with Waymo today, this would suggest that their AI is only at best 10% as effective as Waymo's, and possibly more like 1% or 0.1% or whatever. But since they can't achieve parity, it's not even possible to bound it.
It's entirely possible that their current stack cannot solve the problem of autonomous driving any more than the expert systems of the 60s could do speech translation.
I haven't heard a compelling argument as to why a system that is at best 10% as effective would ever be expected to be the leader.
Data isn’t as useful here as in other domains, since when you change the car’s behavior even a tiny bit, a lot of the timeseries is invalidated. It’s not evergreen, and it can be quite subjective what it means to “pass” a scenario that one previously failed.
Also, Tesla collects data from its fleet, but that data’s fidelity is likely quite limited compared to other companies, because of bandwidth if nothing else. Waymo can easily store every lidar point cloud of every frame of driving.
If you read through the examples on that site, it's actually pretty difficult to find any that have anything to do with ID. There's things like:
- felon (this appears to be far & away the most common)
- moved
- shouldn't have even been allowed to register
- voted twice
- illegally delivered absentee ballot for another person (while not claiming to be that person)
- etc.
The common theme, aside from the absolute rarity of these events (e.g. 36 total in GA over perhaps 40 million votes over ~25 years), is that none of them would be addressed by more stringent identity verification checks at either registration or at the polls.
Clicking through the site, I actually am unable to find a single instance where enhanced ID would have helped. Not saying they aren't in there, just that apparently they are very rare events, even in a dataset of extremely rare events.
He won more votes than that other person who voted. I guess that's only the "majority" when it's convenient for the argument. So you may be technically correct but really, that's the worst kind of correct.
I was inartfully making a claim that voter apathy is a big thing, and as a result no winner can ever really claim to represent a majority of Americans.
IIRC 2020 was the only election of modern times where any candidate garnered more support than "did not vote." Putting it clearly here: neither Obama, Bush, Clinton, Bush, Reagan garnered support of the majority of Americans. It's not a partisan argument I am making here.
A form of government where a third or 40% of the governors routinely fail to exercise their duty has problems.
Without taking away from the reality of housing costs, the long-term chart [1, chart is a year old but looking for trends] suggests other explanations. In particular, the average of first-time buyers was 35, up from 30 as recently as 2010. That's a bad trend.
The overall average appears to be driven by people who have bought homes before. Anecdotally, the Boomers I know bought houses well into retirement where their parents typically made their last home purchase in their 30s. I am less concerned about housing turnover among older people.
A lot of never-purchased-home-before people have money but still don't buy a home. Why? What happens that causes people who have money and can pay to finally buy a home?
I'm asking, what causes people to buy a house for the first time?
For upper middle class people the answer is probably having a baby, so you can see why if people are having babies later in life when their incomes are higher, prices rise.
but the upper middle class is relatively small, depending on how you define upper middle class. Nationally, one might say the 15% who are below the 95th percentile (but locally is what matters for the purpose of buying property, and maybe "upper middle" is 60% to 95% of the local ranking).
> Hacker News handles a huge amount of traffic to dynamic pages
Traffic is not the primary driver of staffing. For a simple app, it is relatively straightforward to use commodity cloud offerings to scale to large volumes of traffic. Even something as simple as Heroku + a CDN can take a small team a long way.
But the NYT is not a simple app. I'm not even willing to accept that their CMS needs could be handled with an off-the-shelf CMS without modification. Without having worked there, I can see:
- CMSs for text/images, audio, video content
- syndication for audio content
- custom? subscription system
- some kind of interface to the printing system
- bespoke game studio
- Web dataviz studio
plus all the stuff needed to run a company as big as the NYT, which will include lots of integrations between things like payroll, accounting, 3P ad networks, reporting, HR software, etc.
I haven't even included the people who might make use of the copious data generated by the business.
These things add up fast.
> Twitter (though not doing well as a business)
That's a heck of a caveat! Most businesses aim to do well as businesses, so current Twitter is not a great model.
> It's still 99.99+% SLA
But this is simply not accurate. 99.99% uptime is < 52m 9.8s annually of downtime. They apparently blew well through that today. Looks like they essentially had the equivalent of 4 years of 99.99% uptime equivalent this evening.
Four nines is so unforgiving that it's almost the case that if people are required to be in the loop at any point during an incident, you will blow the fourth nine for the whole year in a single incident.
Again, I know it's hard. I would not want to be in the space. That fourth nine is really difficult to earn.
In the meanwhile, <hugops> to the Fly team as they work to resolve this (and hopefully get some rest).
reply