ntoshev

Here's what Elon Musk has said about the book:

The only negativity in recent years was due to a book called The PayPal Wars, written by a sycophantic jackass called Eric Jackson. This guy was one notch above an intern at PayPal in the first few years of the company, but gives the impression he was a key player and privy to all the high level discussions. Eric couldn't find a real publisher, so Peter funded Eric to self-publish the book. Since Eric worships Peter, the outcome was obvious - Peter sounds like Mel Gibson in Braveheart and my role is somewhere between negligible and a bad seed. However, to his credit, Peter didn't realize the book would be as bad as it was and apologized to me personally at a Room 9 board meeting at David Sacks's home in LA.

http://m.gawker.com/230076/an-alternate-history-according-to...

I think he means 30-60% of the items listed sell, not 30-60% of the visitors sign up. Not sure how you count this when you expect repeated transactions - perhaps for AirBnb it means properties are booked 30-60% of the time. Liquidity is also about the sellers at the marketplace, not it's owners successfully exiting.

I found the point about (de)centralization of different aspects of the marketplace interesting, can you point me to more advanced resources about marketplaces?

It's a rehash of http://norvig.com/chomsky.html and the talk by Chomsky that led to it (both recent and relevant). I can't wait to read the continuation of that article that Norvig promised in comments.

This is not helpful. Do you know the answer and if so, what is it?

These particular instances are not multi-tenant at this time.

These svbtle blogs render unfocused on iPad 1 in portrait orientation - how annoying.

This was true for the competition I participated in, but it is a considerably weaker statement than grandparent.

How is feature engineering not allowed?

Resources that have been distributed among people in small quantities now can be aggregated and it's a big opportunity. The special case when the resource is people's skills is particularly interesting.

Any links to such projects?

Apparently this site is using CloudFlare, and it's failing:

DNS Resolution Error

You've requested a page on a website (sef.kloninger.com) that is on the CloudFlare network. Unfortunately, CloudFlare is currently unable to resolve your requested domain (sef.kloninger.com). There are two potential causes of this:

Most likely: if the owner just signed up for CloudFlare it can take a few minutes for the website's information to be distributed to our global network. Check back in about 5 minutes and the site should be up and running and enjoying all the benefits of CloudFlare.

Less likely: something is wrong with this site's configuration. Usually this happens when accounts have been signed up with a partner organization (e.g., a hosting provider) and the provider's DNS fails.

Thinking about Google Docs simply as an Office competitor is wrong. If you come to it expecting to do everything you do in Office, it is probably adequate but definitely much worse. However, Docs can do sharing of centralized data, use web APIs, integrate with your mail, and more - things Office can't even hope to match.

The server environment is JVM-based since they expose GWT as the way to build UI. I think I remember seeing Rhino-specific error messages.

The coding environment is not very good, but you can build ERPs, CRMs, document management and similar for small businesses on Google Docs, in a way Microsoft Office / VBA never made possible. The modules it provides (spreadsheets, forms, formatted text document, gmail, calendar, wiki) seem to be very useful building blocks for describing your business processes.

> I was able to automate every part of this except syncing it with my account (since there's no API to access my financial records.)

It's actually possible to simulate a browsing session and have the App Script log in to your broker's site and scrape the data you need. Here is some prototyping code: https://gist.github.com/aaf74e50b9a281396722

Is anyone using QuickCheck as the testing part of a TDD process? Does it work?

There are quickcheck-derived libraries for Javascript, one of them by Douglas Crockford.

Sync deadlines help with at least two things:

- motivation (helps with procrastination)

- discussion (others are on the same stage)

It would be nice if someone installed e4rat on Ubuntu (with ureadahead) and reported the performance.

I would have expected Gmail to have machine learning based filters by now, where you give some examples and it just works.

So, how did Intuit manage to sell their product? What is their distribution strategy?

So these seem to be low stress situations for him, unlike public speaking. YC interviews are obviously high-stress for the founders though.

It's "principle", "principal" means something else.

There is also cpulimit if you want to set hard limits.

They probably mean MW/h ...

Apparently these primitives haven't found their way to the Node.js land yet... and sendfile is still undocumented.

Google Docs storage is $0.25 / GB * year. A random 1TB drive on Newegg was $0.17 / GB. Depending on you utilization and amortization schedule, it may be worth it.

I don't think there were price decreases of the hourly prices for EC2 instances - previous decreases were for traffic, or introducing reserved instances where you get lower price in exchange for commitment.

I think trying to factor in future price decreases in an ec2 cost model doesn't make sense - it is too little and unpredictable to worry about. Don't forget that if you reserve an instance for 3 years and Amazon drops prices, you'd get only partial savings on the hourly fraction of the cost. Even as I use Amazon for hosting (instances running full time) I stay away from the 3 yrs reserved instances because I want to be able to reevaluate my options more frequently.

Like Instapaper, Readability doesn't have an android app, so any other advantage it may have is meaningless for me.

I've tried that. Unfortunately it burned through the battery of my Android phone so quickly that it's unusable unless the phone is plugged in.

I hoped to see a comparison of expressiveness: e.g. a minimal todo list app implemented in several frameworks. I guess this is too much work...

Actually, Google Prediction API is very simple and it covers supervised learning (regression and classification) already. I can imagine very simple extensions (of the API itself, the algorithms would be completely different) to cover a lot of the unsupervised and semi-supervised ground as well.

The algorithms are not disclosed, but the docs hint that they are properly regularized so throwing more features at them is always good.

You still need to be able to reformulate the problems so that they fit a standard ML setting and then know how to tune things, but it looks like the API can get you pretty far.