HNHacker News
TopNewBestAskShowJobs

quizotic

401 karma · joined June 2, 2011

submissionscomments
quizotic··on I'm a black ex-cop, and this is the real truth about race and policing
Too dismissive. I thought the piece was excellent, told stories I hadn't heard before, and made me think about the problem differently.

It's not easy to act correctly in insane situations: http://en.wikipedia.org/wiki/Milgram_experiment. Plus other experiments show that most of us can easily be manipulated into doing horrific things. These are not necessarily rotten police officers. It may be a rotten system.

One nice insight of the post was that modifying the behavior of the few at the top of the hierarchy can effect widespread change. And somehow, we've got to find a way to stop brutalizing and incarcerating so many people.

The more Reddit Hudsons there are, the faster we'll get there.

quizotic··on Vitesse X – A PostgreSQL extension module for faster queries
> Why would some random software be able to make my pg 8 times faster or more?

Peter Boncz (creator of VectorWise, HyPer) says it best here: http://www.tpc.org/tpctc/tpctc2013/slides_and_papers/005.pdf

Remember also that the postgres planner and execution machinery is ANCIENT. Plan scoring still assumes Disk I/O = 10 MemoryAccess = 100CPU instructions. Modern CPUs are SIMD and Vector processing, but PG planning/exec machinery isn't vector aware. There's a 200X difference between L1 cache and main memory access that the PG planner can't see. And what about SSDs and NVMe vs disk? Spark gains its speed through caching intermediate result sets along with the ability to reconstruct them. But postgres doesn't cache intermediate results or query plans last I checked. Plus a million other things.

Postgres is wonderful, but not because it has natively fast analytics. 8X is nothing. 180X is better, but not as fast as Vertica, Aster, Greenplum, et al. I say kudos to Citus and Vitesse for leading the way.

quizotic··on Vitesse X – A PostgreSQL extension module for faster queries
probably not snake oil; and 180x faster on tpch-1 is no where near physics speed. but these guys are just starting out. i'm sure they will go even faster once they pull out the stops.

the nice thing about their model is that it's a pure accelerator play. plug it in and go faster, pull it out, or replace it with something else. in theory, there should be no change to the application/query layer. It should be as simple as entering "create/drop extension vitessedata;"

quizotic··on Ask HN: I'm terrified of death. What do I do?
And here's a trick. It works for me, but not others. Einstein and physics say we live in a space-time continuum. The past and the future are just different points, different places.

What you've done already, where you've been, what you've felt, these constellations in space-time are eternal. There's nothing that can erase them. They happened, and nothing that happens in the future can change that.

Your five year old self lasts forever, but is also gone forever. So is your nine year old self. You are immortal and your are ephemeral. Both, at every moment. In that light, death is nothing different at all. You've been doing it just fine all this time.

quizotic··on Ask HN: I'm terrified of death. What do I do?
When my wife died, death lost all of its teeth for me. It's not worth the price, but losing someone you love pretty much solves the fear thing. The valley of the shadow is a scary rollercoaster on the way in. But on the way out, it's just barren and lonely. Not scary at all.

More practically, there are some tricks that can make this fear go away. I've seen experienced psychotherapists help people with this quickly. If you can afford a couple of sessions and don't mind the stigma, you might get some quick, long lasting relief.

quizotic··on Ask HN: Programmers Above 50, Is It Possible to Have a Career Past Your 50s?
I'm 59 and based in New England. I worry about this all the time. But so far, the worry has been a bigger problem than the reality: I've left stable employment situations to start up my own things, thinking that was my best insurance against hypothetical age discrimination.
quizotic··on Secret Shuts Down
One reason is when the investors own > 50%, control the board, and decide to recoup 10 cents on their dollar.

But I've known entrepreneurs who made the decision independently, feeling it was the right thing to do.

And it is the right thing to do if the company has no other prospects or the team is falling apart.

But even so, returning the balance just keeps some salt from the wound. It doesn't make anybody happy.

quizotic··on Woman behind Pakistan’s first hackathon, Sabeen Mahmud, shot dead
Oh, that's so sad!
quizotic··on Amazon Finally Discloses Cloud Services Sales, Showing 49% Jump
Does anyone have insight into AWS revenue per service? Does EC2 generate the bulk of the revenue? Is Lambda the fastest growth area? How is Redshift doing?
quizotic··on Sick time is logarithmic
I looked at sick time a couple of years back and noticed that when a frequently sick person stayed home sick, another frequently sick person followed suit within a day or two. I wonder if anybody has a Pearson correlation on that.
quizotic··on ZeroDB – A Peek Under the Hood
Actually, CryptDB does (or at least did originally) use homomorphic encryption (HE) for evaluating SQL arithmetic expressions. While the earliest suggestions for HE were terribly slow, CryptDB found one that was practical. If memory serves, there have since been HE improvements claimed by IBM and/or Google.

But as michwill states (and CryptDB acknowledges), their approach either leaks information or precludes classes of expressions. For example, their order-preserving encryption could be susceptible to a brute-force attack via 'a > b' ... '(a+1) > b' ... '(a+2) > b'

quizotic··on MongoDB Is Raising Another $100M
Not stupid, but there are different perspectives that reduce the scariness.

With are relational model, your working set needs to be joined together from pieces, so you want ACID to ensure that everyone sees a consistent set of pieces.

But with a document model, you can 'pre-join' your working set into a single object that has everything you need. And that object doesn't have a fixed schema, so it can grow and evolve over time.

While Mongo doesn't provide ACID across documents, it does guarantee atomicity, consistency, isolation and eventual durability for SINGLE documents.

IF your application can live with a universe that consists of a single, arbitrarily complex object - then Mongo is as within epsilon of being as safe as a regular ACID transaction system.

quizotic··on Do elite software developers exist?
I wonder if we're evolving scoring systems right now. Not so much with hackathons and sites like codefights.com but through data mining github. I'm thinking of a recent post to HN in which successful github projects were correlated with a contributor who added more code than all other contributors combined. I suppose if I were looking for a 10X developer, I might reach out to people like that?
quizotic··on Some Thoughts on Founder Liquidity
> fuck Sam Altman on this one

Really? Most early investors make 90+ % of their return from the tiny fraction of investments that become huge. If the founder is willing to cash out for current market value, doesn't that imply the founder doesn't believe the equity will quickly grow 1000-fold? And if the founder doesn't think the company will become huge, wouldn't the investor be foolish to think so?

This isn't about squeezing the founders, it's about passing on stuff that might be glittery, but probably isn't the mother lode.

quizotic··on Ask HN: Done with startups. Contemplating freelancing
I'll second this. In my case, fear of famine (which never actually happened) often led me to take on too much work to have a side project. Sometimes, I couldn't do as good a job on concurrent projects, but was afraid to do things sequentially or turn down projects.

FWIW, the business and sales skills tptacek talks about end up helping when the startup bug strikes again ... which it probably will.

It's been said elsewhere, but one of the most important things is to hold firm on price, probably a much higher price than feels reasonable coming from startups. If you're not charging by the project, charge by day, not by hour. You have extra costs as a freelancer, and you provide extra value. So build that in.

quizotic··on US Senate Report on CIA Detention and Interrogation Program
I'm in total agreement with your sentiment and those (so far) in this thread. But I want to question the pretext of the interrogations: the assumption that Americans will support any action that keeps America safe, the "safety first" assumption.

While the importance of safety is reflected in bulk of our public discretionary expenditures, Americans have other values. Americans value liberty. Americans value humanity. Many value these above safety (e.g. N.H. state motto). Sadly most people I know choose safety first and close their eyes to the corollaries, which leads us to where we are now.

quizotic··on If Eventual Consistency Seems Hard, Wait Till You Try MVCC
One of the functions of an Operating System is to provide (the illusion of) isolation between Processes and the resources they acquire. Roughly,

Transaction : Database :: Process : OperatingSystem

The four transaction isolation levels, defined nearly fifty years ago, were an attempt at categorizing incomplete isolation. Can you imagine launching an operating system process and saying "It's OK if my memory locations are changed by another process" Or "Don't let my memory locations be changed by another process, but it's OK if another process prevents me from deleting a resource". That's what we're asking of our non-serializable database transactions.

Why not just provide strong isolation? Partly because it is hard and partly because the performance impact of strong transaction isolation is greater than the performance impact of process context switching.

But if you're living with less than strong transaction isolation, then tautologically, strange and unexpected things eventually happen: seeing state that never existed, seeing state changes that you didn't make, failing to see state that you should have seen. Rarely, the application can reliably detect and handle some of these situations. Typically, the application only thinks it can.

Sometimes, I think the core issue is with the notions of isolation and serializability themselves. Developers want to believe that events are noticed simultaneously with their occurrence, and that all observers (transactions) see the same history unfold in the same order. But the pesky physical world doesn't work that way.

quizotic··on How to Be Intoxicated
"overwrought and pretentious?" I felt jealous for the deep insights and the ability to communicate them so clearly. This was beautiful thinking and beautiful writing.

Perhaps Euripides is an even better thinker and writer. But I wouldn't have known it without the author's explanation.

quizotic··on Confessions of a mortician
Funny. Poignant. A little scary. A little gross. Eye opening (a la the Matrix). That's a lot of value for a 10-minute read.
quizotic··on Data structures and Algorithms in C/C++11
for example: http://stackoverflow.com/questions/461203/when-to-use-virtua...

class B : public A { ... } void foo() { A* someA = new B(...); ... delete someA; }

if A destructor is not virtual, then B's resources leak.

quizotic··on How to land your first CTO role?
I became VP of Engineering in a pivot/refinancing. Later became CTO in another pivot/refinancing. Subsequently was offered both positions several times, mostly when things were not going well, occasionally when just starting out.

The CTO position is usually more outward-facing, with some limited ability to affect change internally. As CTO, it's useful to blog, speak-out, meet prospects, acting as a thought leader and technical evangelist. Some % of time can be devoted to future product concepts. As CTO, you would be expected to be familiar with your technology ecosystem and have a good sense of its tectonic forces and near term direction. If tea leaves need reading, your opinion would count heavily.

The CTO role can be tricky to navigate. The easiest path internally is to be a resource to engineering, marketing and sales, offering to help each function, and communicating frequently. Dangers include being perceived as (1) sending messages not coordinated with marketing (2) stealing engineering resources (3) setting product/company directions independently.

Good luck. It's a fun position ... if you don't mind not having control.

quizotic··on How to Start a Startup – Lecture 14: How to Operate
+1 Track complementary KPIs, so as you go up on one, you don't go down too far on the other

+1 Grade/compensate based on ratio of output divided by resources, so that managers don't have incentive to create empires.

+1 Calendar audit vs stated priorities

+1 Company must have one voice and a simple message

quizotic··on Introducing Modelling by Example
Does anyone know what this means? Can you explain it to an old C++ hacker? Near the end, it looks like "ubiquitous language" (UL) is being used to generate API function signatures. Is that how UL connects the business reqs to the implementation?
quizotic··on Never Negotiate Your Freelance Rate
Chacun a son gout... but I agree with OP. When I've been asked to discount, I recommend two or three other consultants. I sell hard for them, extolling their work, offering to make an introduction. Someday, a prospect will take me up, I'll lose some business, and win a favor with a colleague. But it hasn't happened yet.
quizotic··on Why GNU grep is fast (2010)
> Make it do less.

Yes, but less in a particular way. It's important to skip processing stuff that can't matter in the end. In this example, skipping over bytes that can't matter because the end doesn't match. In DBMS, it's important to skip over records/columns/values that can't contribute to the answer.

The surprising realization in optimization is that procrastination is a virtue. Lazy beats eager.

quizotic··on Three reasons why the Semantic Web has failed
Yeah, the article is crap. But did you take a look at the site: bottlenose.com? It's kinda gorgeous.

Having been in the biz (NLP search with sentiment extraction) once, it still comes down to precision and recall, with poor precision being the fastest way to lose sales. I didn't see any indication of bottlenose accuracy. But maybe I was too blinded by the beautiful d3.js.

Speed is also an issue. Machine Learning takes setup time and good corpus sets, after which it's pretty fast. Traditional NLP is faster to start but slower and less accurate after. Neither is remotely close to real-time, which makes me wonder what they're really delivering.

Also, at least in my experience, I couldn't get product managers/marketers to give a hoot. But ad-agencies ate it up. And boutique survey shops. And sometimes CEOs. And sometimes customer-service organizations that had too much inbound hate mail and had to triage.

I think they may be smoking it to think they're going to get inbound traffic. I had to pound doors.

quizotic··on Why You Should Never Use MongoDB
Funny thread!

MongoDB's query performance is typically sub millisecond, and rarely as much as 5ms. I don't think you can through the postgres/mysql parser in 5ms, much less optimizer, planner, and execution stack. Couple that speed with a dead simple API, and you've got a thing of beauty.

So yeah, if you don't care about ms latencies, if you've got a fixed rectangular schema, if want to write queries with lots of joins, if you need ACID guarantees ... etc ... then by all means pick your favorite RDBMS.

OTOH, when you need to manipulate relatively self-contained objects quickly, when those objects don't have a fixed schema, when availability is more important to you than immediate consistency, then why would you choose anything other than MongoDB?

quizotic··on Ask HN: YC rejected. Now an acquisition scenario came up. Help?
1. Why are you still mourning YC'14? 2. Why are you asking for uninformed advice on a public forum?

In situations like these, I like the Mathew Looks rule.

Mathew was a USCF expert on my high school chess team, trying to teach me to be a better player. He set up a chess position with chances for both sides, and asked which side I liked better. "I have no idea," I replied. "That's why you lose. It doesn't matter whether you're right or wrong. It matters that you have an opinion."

Form your own opinion about this situation. Like it. Feel confident about it. Then follow through, as well you can.

It doesn't matter what anyone else thinks. It matters what you think.

quizotic··on Animate.css
Very Nice!
quizotic··on Processing 6 billion records in 4 seconds on a home PC
The two most interesting things about this article to me were unstated.

1. The TPC-H benchmark is measured in price-for-performance ($/QphH, or dollars per queries-per-hour). At 4 seconds for Q6, he's getting ~900 queries per hour. The cost of his rig is probably ~$2k, so he's under $2 per QphH. The top TPC-H scores are around $.10, but <$10 is pretty good for a first go.

2. The standard knock against GPU processing is the time it takes to load GPU memory. GPU processing may be blazing once data is in memory. But there was an MIT paper last year claiming you couldn't load the GPU fast enough to keep up. Evidently, he's keeping up.

With regard to comparing his performance to hadoop/hive - yeah it's apples and oranges, but he's in good company. Hadapt, Hortonworks Stinger, Cloudera Impala, Spark/Shark and others all rate themselves on how many times faster they are than Hive.

And frankly, I don't buy the whole "the point of MR is for huge, horizontally scaling networks" If you factor out Yahoo!, Facebook, Amazon, LinkedIn and a few others, the largest remaining hadoop clusters are all WELL south of 1000 nodes. And most run on homogenous high-end hardware.

← PreviousPage 4 of 5Next →