1,645 karma · joined March 11, 2018
i.e.- this was not a rerquest or taner saner to provide context, but a request for @ericdiao to provide context.
One of the nice things about HTTP(S) is it has rerdirect semantics (to shed load) and requests are easy to send through a load balancer. 30k requests per second for static web data is well within the capability of a modestly scaled cloud or random assortment of machines in someone's data-center.
Also remember old networks that fed mainframe apps used to be pretty slow. I'm more familiar with old school travel agent tech. it was not uncommon to have offices of 5-15 travel agents all sharing a single 56 kilobit data line. So much of the coolness of mainframes was being able to aggregate hundreds or thousands or tens of thousands of remote terminals in the time when a 1200 baud modem was considered "fast."
And when they say "transactions" they're not talking about web requests, they're talking about multi-phase commit over multiple sub transactions, all of which have to be unrollable if any of them fails.
When I was at IBM/AIX division in the early 90s, we were excited we could get 30 tpmC out of a $15k RS/6000. The AS/400 team down the hall would snicker at us and invite us to watch their $30k entry level machines doing something closer to 500 transactions per second.
But the thing is... they were using a much simpler data representation scheme that made it MUCH easier for data inconsistencies to creep into the process.
So... modern PCs are more architecturally similar to old mainframes than old 8/16 bit micros. And they're certainly faster. But those old systems had plenty of tricks up their metaphorical sleeves.
I'm not sure this kind of misbehaviour reflects well on our brand.
Do you have a contact at the university I can talk to?
But the OP does have a point, they each can introduce more trouble then they're worth. Were I to write this post, I would have titled it something more like "Systems Ideas Yo Really Should Think About Long And Hard Before Doing."
But yeah, that might not be enough warning.
Seems to me the answer to 'Can AI do maths yet?' depends on what you call AI and what you call maths. Our old departmental VAX running at a handfull of megahertz could do some very clever symbol manipulation on binomials and if you gave it a few seconds, it could even do something like theorum proving via proto-prolog. Neither are anywhere close to the glorious GAI future we hope to sell to industry and government, but it seems worth considering how they're different, why they worked, and whether there's room for some hybrid approach. Do LLMs need to know how to do math if they know how to write Prolog or Coc statements that can do interesting things?
I've heard people say they want to build software that emulates (simulates?) how humans do arithmetic, but ask a human to add anything bigger than two digit numbers and the first thing they do is reach for a calculator.
There are a couple things going on here:
1. DSLs aren't "bad." But they may require more forethought than you typically have had to apply to typical programming tasks.
2. Doesn't perl6 do something similar? It was about the only thing about perl6 I liked. Insert reference to your favourite dynamic grammar system: icon? forth? some lisps?
3. something that is sorta new to think about is SQL is supposed to be a declarative language and behind the scenes there's a planner that knows what to do to put a particular record in a particular state. And yeah, you're doing something similar, changing the semantic rules to produce an AST, which you're still using with previously coded code to determine the semantics of the thing you wrote in the new grammar. But that's essentially what the OP said here.
4. I agree with the author that maybe PEGs aren't the most awesome thing in the world, but they seem to be well understood and actually doing something is better than trying to mzke things perfect.
5. I liked the author's write-up, but as an old programmer take umbrage at the idea that changing your parser in the middle of a program is "crazy", we used to do this... well maybe not all the time... but with a greater frequency than we do today.
Should we have dropped the bomb?
That's the last decision in a series of policy questions on both sides, each a complex response to complex questions going back at least a decade.
FWIW. You can see the fourth gadget at the National Museum of Nuclear Science & History in Albuquerque.
If your assertion is people get a STEM degree and then don't want to work for a scammy startup that has no plans for profitability and funds operations by repeatedly returning to the VC money-trough. That's not a failure. That's a success. Students have learned how to apply critical evaluation skills to avoid hype-laden zombie startups pitching employees on equity grants that will never be valuable (much less lucrative.) Instead they're moving into careers related to thier degrees: biology, geology, mechanical engineering, etc.
STEM does not mean "I copy code fragments from Stack Exchange and Co-Pilot all day."
I happen to know a few googlers socially from my days in SFLUG and BayLUG, so I can sometimes get a note to the appropriate product manager or engineer through the friends-of-a-friend network. But going through the front door has never worked.
I believe that Google Maps is perfectly fine in most of the bay area. It's generally acceptable for roads that a google street view car has driven down. But pretty much a lost cause for other roads.
Corner me at a conference sometime and I'll tell you about how google maps sent us four-wheeling through eastern california fire roads (a dirt road that collapsed after we turned around at a critical junction) or told me to get on and off the highway the first time I drove from Tacoma to Seattle (what should have been a simple 45 min drive with traffic turned out to be a 2 hour slog fest because I didn't have enough local knowledge to realize maps was full of itself.)
While it's great they no-doubt are providing an internship for two or three PhD candidates, I think they may want to fix their data before thinking AI will improve the experience.
The AI does not know the data you're training it with is garbage. If yoy do it right, you may be able to spot anomolies in the data or auto-cluster bits of data, but if you train any sort of CNN on garbage data, you're going to get garbage out.
So while this may be great for people commuting from the GooglePlex in Mountain View to the Google facilities in San Francisco, and it //may// help people traveling along I-5 in California, I fear the garbage aspect of their geo data will not be magically solved by adding AI.
So instead of leaving a snarky comment that gets @dang mad at me, I'll just say I'm hopeful we're at an inflection point for Boeing. I really want to love the company again, they've made it hard to love some of the decisions they've made the last few years (decades?)
I don't understand Hacker News sometimes.
[note, however, as a conniseur of fine s*rk, I did not downvote you.]