If e.g. taxes are raised, and GDP decreases, it is theoretically possible that net revenues decrease even though tax rates have increased. Or, revenues may increase but the debt to GDP ratio still increases (because the lower GDP) and borrowing costs increase as a result.
Note that this means, to me, we need to get our shit together sooner instead of later. A little austerity now may prevent outright default later when significant austerity is forced.
The most effective way is likely not broad-based cuts but targeted cuts to old-age spending via means-testing to induce liquidation of 401(k)s or other savings or tangible property to pay for their own care. I feel the longer (and slower) this trainwreck proceeds, the more likely this outcome is palatable. Roughly 52% of retirees have 250k or more in savings [1]. It is far preferable to drawdown those prior to either cutting benefits or raising taxes.
Fyi asset testing and means testing are different, but yes that is absolutely a good way to implement things from a utilitarian perspective. From a political perspective... Old folks vote. It is a tough situation.
Assets are for practical purposes means and when relatively comfortable retirees draw on SS/401(k)s it is income. There are broadly two choices for the wealthy retirees:
1. cut back on SS and other benefits you can afford out of pocket
2. tax your income higher
I would wager they would choose 1. To this group, the taxes hurt worse than the benefit losses
For poor retirees:
1. cut benefits
2. don't cut benefits
I would wager they would choose 1
The other two commonly suggested fixes -- tax the wealthy and broad tax increases -- also fail in my opinion. The holes we are talking about are in the hundreds of billions per year and won't be fixed by taxing just the wealthy. A broad middle class tax increase will also likely be rejected as unpopular.
It really is my belief that a slow-motion failure here is best. If any fix occurs as the social security trust fund is at 0 cash and while the bond market is in revolt, there will be no fiscal room to maneuver to raise cash from the debt which forces rationing.
In Commonwealth countries, it's possible to initiate a private prosecution (by a person who hasn't received criminal justice). These don't usually get very far, but I honestly wish these could be used to prosecute people who are treated differently for the same crime because of their status, or at least set a precedent that such activities are definitively okay.
> In Commonwealth countries, it's possible to initiate a private prosecution (by a person who hasn't received criminal justice). These don't usually get very far
> Between 1999 and 2015, more than 900 subpostmasters were wrongfully convicted of theft, fraud and false accounting based on faulty Horizon data, with about 700 of these prosecutions carried out by the Post Office. Other subpostmasters were prosecuted but not convicted, forced to cover illusory shortfalls caused by Horizon with their own money, or had their contracts terminated. The court cases, criminal convictions, imprisonments, loss of livelihoods and homes, debts, and bankruptcies led to stress, illness and family breakdowns, and were linked to at least thirteen suicides.
Hundreds of innocent people went to prison, were financially ruined, thirteen took their own life, because a well-funded entity found it easier to privately prosecute its contractors than to question its software vendor. And the court system let them do it.
For some insane reason I am unable to comprehend, the UK government still allows Fujitsu to bid on government software contracts, despite being responsible for all of the false imprisonments, trauma, and tragedy you laid you in your post.
I have a feeling Fujitsu took the blame for mistakes done by the Post Office in order to get future contracts. The contracts are padded to pay fines and damages.
I understand the government would have difficulty excluding Fujitsu under its own procurement regulations at the time (Public Contracts Regulations 2015). Fujitsu voluntarily withdrew its bids in 2024. The goverment passed the Procurement Act 2023 which came into force in February 2025 and there is now a national debarment list, though I note Fujitsu isn't on it.
Also, it's important to note that Fujitsu didn't prosecute anyone, they just lied to the Post Office, lied to subpostmasters, remotely altered evidence, gave false testimony in court, etc. It was the Post Office itself that prosecuted its own subpostmasters.
In Australia at least, the Director of Public Prosecutions has the power to take over the matter, and they often do.
They can then do one of two things. They can either continue running it if it’s in the interests of justice to do so (public interest, reasonable prospects of success etc), or they can shut it down, as they tend to do for sovereign citizen nonsense and the like.
Admittedly a controversial take, but I think some of the AI skepticism I see is driven by status anxiety, alongside the perfectly legitimate complaints about verbosity, low-effort output, lack of a human voice, etc.
The bar for what counts as expertise is shifting. If someone with relatively little background can use an LLM to get to a reasonably informed answer much faster than before, then experience and credentials alone carry a bit less signalling power than they used to. That obviously doesn't mean the inexperienced person suddenly has the judgement, context or intuition of an actual expert, but I do think some of the hostility toward AI comes from discomfort with that change.
AI is still wrong, goes in weird circular loops, and doesn't solve the fundamental human problems around judgement, organization, learning curves, accountability, etc. But it does seem to be raising the baseline expectation for what a competent knowledge worker should produce, and I feel that shift is genuinely uncomfortable for a lot of people who based their identities around being rigorous abstract problem solvers.
Ironically, some of the AI 'trustism'[1] is very much driven by anxiety the fear of being left behind if one does not jump on the train.
I agree that it shifts the notion of what is expertise, but for the worst. It's very hard to tell who is actually an expert, and who is mimicking the expert.
The consequences vary from annoyance to downright catastrophe.
e.g.: a confidently wrong medical advice
On the other side, sometimes the expertise is not a requirement for the user: "If it works and does what it says on the box, that's all I care".
I like the fact that you use the word "some", awesome to see some nuance. It's a tricky situation.
---
[1] I am being an idiot on purpose, I think the word skepticism was completely derailed from its origin and meaning.
A minimum amount of skepticism is very healthy, especially for software engineering.
Most software engineers, and science-based professions for that matter, are skeptics at heart. And it's a good thing.
Skepticism is the base of science; science thrived when we admitted ignorance and became skeptic towards our own knowledge.
>3.7 Flash is available through the end of the year at an introductory price
1 of $0.75/1M input tokens and $3.75/1M output tokens. This price combined with the enhanced model performance enables developers and customers to scale production-ready agents cost effectively.
Introductory pricing until December 2026 implies no significant Gemini Flash developments until the next year.
I think it's just meant to make it more competitive, Gemini has kinda been behind in everything except maybe multimodal. It's only 3 weeks after Flash 3.6, so if they really wanted to, they could probably do a 3.8 Flash before then.
3.5 Pro was supposed to be around the corner two months ago. 4.0 Pro is some ways out as they recently stated they are seeing some promising early results from training. It didn't sound like a release is imminent.
My university let us take 3-4 courses, up to 12 credits total, as Pass/Fail, but only for electives.
I ended up taking just one that way. The policy was really designed for the breadth year because in upper years, you had much less room for electives, but you could try a difficult or unfamiliar subject without worrying too much about your GPA. If you discovered you liked it enough to switch into that major, the course would then be converted to a regular graded course.
I think allowing 3-4 Pass/Fail courses in the first year strikes a good balance. It gives students room to challenge themselves and explore unfamiliar subjects, while still preserving grades as a useful signal of whether they're prepared for college-level work.
Is the assumption here that inference costs will stay roughly static, or that frontier models will keep getting more expensive quickly enough to offset efficiency gains?
Because I don’t think “the subsidization train is going to end” necessarily means current pricing becomes impossible.
If capital keeps pouring into frontier AI, companies still have an incentive to subsidize access while competing for users and market share. And if that subsidization starts drying up, there’s even more incentive to bring inference costs down by making smaller and cheaper models catch up to today’s frontier capabilities.
So either way, I’m not sure you can extrapolate from the cost of serving current frontier models to what equivalent capability will cost two years from now.
I think like you mentioned, the practical reasons are disproportionately oriented around either privacy (, a clear constrained workload (need to OCR files/transcribe audio, and there isn't really a clear or meaningful reason to switch out the model to chase new incremental gains), or regulatory compliance (e.g. source code, patient data, can't leave the country).
Obvious caveat being that Claude Code might give you a confident statistic without being able to validate if the question can be validated with the sample (StatsCan suppresses C grade data and below).
I don't think this would make a big difference for the threat model described in the OP? They'd still be able to use your IP Address and potentially do nefarious things through your role as an unwitting proxy.
Using one device as a proxy is a few steps away from trying to exploit and infiltrate the other devices on your machine as well. An unwitting proxy is already crossing the line to putting a fox in the henhouse.
Limiting what outbound access devices can/can't have is an important skill to learn.
I think from a technical perspective, newsletters are a solved problems with lots of options as you've mentioned.
Substack thrives because it's built a reputation for having well-reasoned or engaging voices in a newsletter format. They pushed a big campaign to onboard writers like Bill Bishop (Sinocism) onto the platform early on. Because of that, newer writers are likely seeking to tap into the network effects/discovery systems that comes with having big names on the platform.
People log onto Instagram, Tik Tok, X, etc. because the people they're interested are already on the platform, and the platform provides a recommendation system that allows newer "creators" to tap into an existing audience for others in their niche.
reply