Anyone found a good file upload solution for MCP? Or is the best practice to use HTTP to upload files outside MCP?
643 karma · joined February 13, 2014
> ... Despite this, there were no deaths caused by acute radiation syndrome. Given the uncertain health effects of low-dose radiation, cancer deaths cannot be ruled out. ...Predicted future cancer deaths due to accumulated radiation exposures in the population living near Fukushima have ranged in the academic literature from none to hundreds.
> ...For evacuation, the estimated number of deaths during and immediately after transit ranges from 34 to "greater than 50". The victims include hospital inpatients and elderly people at nursing facilities who died from causes such as hypothermia, deterioration of underlying medical problems, and dehydration. The old people and already sick, were more likely to be injured because of being relocated than damaged by radiation.
* https://en.wikipedia.org/wiki/Fukushima_nuclear_accident_cas...
Yeah, good point. At a cultural level, this was a change from an Oral world where _knowing_ what to recall, being able to, and being 'witty' about it mattered a great deal; to a Literate world where things could be cited, 'looked up', and compared.
Unlike Oral cultures where a point of view can be questioned and held accountable, a written version changes these expectations since it can only be conveyed one-way, i.e a trust-me mentality.
Secondly, humans having evolved in an oral world are used to much greater exertions and development of their memory (eg: recalling and narrating complex oral epics) which writing (and later computers) greatly weaken.
Ong makes a similar argument in Orality and Literacy (1982) https://www.goodreads.com/quotes/7182447-most-persons-are-su...
> Pure mathematics is dead. Long live mathematics. I think all of interesting mathematics is applied mathematics in the end. Powerful AI means that the level at which we can do applied mathematics will be so much higher, though, and many more people will be able to be "mathematicians". The importance of pure mathematics is often argued for by citing examples of important applications that used pure mathematics invented a long time before the application became apparent. We can reverse this argument: by properly developing the mathematics our applications need, we surely will obtain all of interesting pure mathematics.
Perhaps the pace of applied mathematics would rise sharply, given cheap intelligence. And this* may end up being the forefront driving progress in mathematics.
*Or maybe a split between the human domain and the practical real world. Where the human domain might end up with a variation of a "No machine contributions" policy. Sorta like the recent gcc policy.
Technically, he is not. He returns in the final scene.
True but this can be a lot larger, maybe even a 1 exabyte instruction size since it depends on he programming language used to express the program. And even accounting for the invariance of runtimes at scale*, the constant factor might be gigantic since we have to build up a lot of cellular machinery first. Reminds me of the Carl Sagan quote: To make a sandwich,you must first construct the universe!
* For example, to translate a program from language A to language B, you can dedicate a constant size to write a language A to B translator. Thus at large program sizes , the kolmogorov complexity (i.e instruction set size) is fairly similar between programs since the size of the program dominates the size of the translator program which is constant. But the constant factor for cellular machinery might be gigantic**
** Or not if we only need a rough simulation
> Zuck is also revealed to have given the Chinese state access to all of Facebook
Tbf, the book actually makes the right claim that it's Chinese user data, not all of Facebook so the article is to blame.
Asking since we've managed to increase superconductor temperature several times in the past, right? (to ~ -130 degrees celsius right now IIRC). Why is our current temperature of, say ~30 degrees celsius special?
> In a highly popular statement, we are told that the family has progressed from institution to companionship. But, as Ortega y Gasset has written, “people do not live together merely to be together. They live together to do something together”. To suppose that the present family, or any other group, can perpetually vitalize itself through some indwelling affectional tie, in the absence of concrete, perceived functions, is like supposing that the comradely ties of mutual aid which grow up incidentally in a military unit will along outlast a condition in which war is plainly and irrevocably banished . Applied to the family, the argument suggests that affection and personality cultivation can somehow exist in a social vacuum, unsupported by the determining goals and ideals of economic and political society.
Going on a tangent, my current beliefs are that:
1. Social functions (i.e accomplished through association) has always had, and will always have high marginal utility, independent of and utilising any technology.
2. That there are political and not technological barriers suppressing it in our current age.
3. That humans are evolved to interact with large numbers of humans (probably seasonality), and that our evolved sociality is scalable even to the present day and beyond (i.e a rejection of Dunbar's number as an evolved constraint)
The second component, frequency of change is equally important as when faced with tradeoffs, we can push high cognitive load to components edited less frequently (eg: lower down the stack) in exchange for lower cognitive load in the most frequently edited components.
Honestly, some form of digital credits work just as well here IMO (barring regulatory hurdles). No crypto necessary.
Tl;dr* He claims ads were the original sin of the web, built for a human internet. For the coming agentic web where most browsing would be done by agents, he proposes a new protocol that has payments integrated into it. Specifically agents paying for accessing content using micropayments with stablecoins.
Also interesting to note that this is one of the most submitted articles of Stratechery to HN but gained no traction. Latest submission: https://news.ycombinator.com/item?id=44073241
* A summary can also be found here: https://stratechery.com/2025/the-upheaval-coming-for-the-int...
> Paid competitors cannot compete because people won't pay.
Seems to me they're different groups of people. Otherwise, if people hate ads, why don't they pay?
I don't agree it's an everyone loses situation. Enough people don't mind or prefer the ads based ecosystem for it to continue.
IMO, more sophisticated ad blockers (I.e real control over one's devices) used by more people, would make the ad tech ecosystem a lot less profitable. After all, ads only work when you can control what the other person sees.
I'd tried applying memory training lessons from this book a few years ago and written about my experience: http://web.archive.org/web/20210301185111/https://ppsreejith...
1 Minute Creation Video: https://youtu.be/dANHVKezZ4w
Also, I remember transferring lots of data through the bridge could be a bottleneck for some use cases. Is that effectively solved with this architecture?
IME benchmarks, though valuable, don't fully reflect the real world, often only reflecting the easily quantifiable. The best way is to be able to quickly try out an agent to see how it performs on your work environment. Sort of like having a private test set you can try different agents on to see how they perform in the real world quickly.
Disclaimer: I'm building MinusX, a data science agent (github.com/minusxai/minusx)
This really depends on whether your tool can handle the scale. We only use a sample of the outputs when constructing the context for your instruction so it should be independent of the scale of the data. We mostly use metadata such as table names, fields, schemas etc to construct the context.
> Also does this work with self hosted Metabase or Metabase Cloud? Or both?
Yep, it should work on both :) We have users across both