Astronautics is a problem! I'm interested in search engines, and keep hearing this hype about peer-to-peer searching. Napster uses a central index -- what's so peer about that? Gnutella doesn't, but it's very limited in what you can find: if your local group isn't interested in folk music or XML DTDs, you seem to be out of luck. That's not a very revolutionary search engine to me.

The only person who I've heard speak passionately about peer and distributed search engines is, as you might expect, a recently-retired AOL Fellow, CEO of a company they bought several years ago. Fits the Architecture Astronaut stereotype pretty well (though he's a great guy and dresses well).

-- Anonymous, April 17, 2001

Answers

To be fair, Dave *did* plug this article on Scripting News.

-- Anonymous, April 20, 2001

Thank you! This is something I've been yelling about for years, so it's great to see someone else talking about it too.

Think about it: when was the last time that you said "Hey, I'm going to go buy a hammer today, 'cause I saw an ad for a neat new hammer." Nope, doesn't happen. What does happen is that you need to build something, and if the hammer you have doesn't do the trick, you go out and buy one that does. But the project always, always comes first, not the tool.

This is why I laughed at peer-to-peer, and push technology, and OpenDoc, and another dozen technologies. Real people rarely look at tools and say "Hey, I could build cool stuff with that, so I'lll buy it," they say "I want to build cool stuff--now, what's available that'll help me build it?"

-- Anonymous, April 17, 2001

Ease up, folks. I'm pretty sure you're not seeing these "high" ideas on TV or in the newspaper. I tend to see them some trade rag like eWeek, or Object Broker Magazine, or the Wire Protocol Gazette. That is, I see them touted right where they're supposed to be touted.

With that said, there are some esoteria which aren't anything special at all. (XML is -not- going to make your browser understand ticket sales; instead, you'll have to have a jillion plug-ins to understand a jillion businesses' mutually incompatible XML specs until the standards wars die down.) But some are. If the advertisers are doing their job, these ideas will only make it to the audiences most likely to know something about them. If you don't, then don't sweat it too much. (Either that, or you should suddenly realize you've fallen behind in your field, which happens at times...)

-- Anonymous, April 17, 2001

Ease up, folks. I'm pretty sure you're not seeing these "high" ideas on TV or in the newspaper.

Here's an example. I think the NY Times counts as a non-tech publication; don't you? And SOAP and XML-RPC are tools, not end projects.

And I've seen plenty of non-tech industry writings about .NET and Hailstorm, that's for sure.

-- Anonymous, April 17, 2001

What are those photos?

I recognize my alma mater in the first photo, and I'm guessing a comment was being made about it. Fair enough. But what do the two other photos represent?

-- Anonymous, April 18, 2001

Those that live in pressurised suits isolated from the rest of the universe aren't confined to big industry wide acronymic architectures (though an acronym is one of the fastest ways to spot them). Just about any project worth thinking about has a tendency to latch onto a single pattern and try and replicate it throughout the development.

I admit I've been guilty of this in the past. It is extremely seductive to look at a problem, discern the pattern and then leap to the assumption that solving that pattern will fix everything else. (I'm trying hard not to use the word paradigm here).

There _are_ patterns that are useful, having all primary keys the same length and using the same algorithm to generate them is a reasonable thing to do. But it isn't always the right thing to do for the data. If the data is very small, but numerous then adding meta data to get to a particular row is unlikely to be a good thing, the ratio of hit to size of data and key is likely to be uneconomic and so in that case the data itself is likely the best key to have, hashed or otherwise.

Learning when not to apply a pattern is as important as spotting patterns in the first place. Very quickly, what becomes a simple 'oh this is the same as that, its just different in one regard, lets create a flag', turns into a maintenance nightmare where the code is broken because a special test in a friend function just invalidated the logic in the tree for the general case.

-- Anonymous, April 18, 2001

I've met a couple of these types. They tend to be balding or silverhaired techies who have spent too many years working with the marketing organization.

One of these guys like to talk about the difference between "models" and "paradigms" - his revolutionary idea was "communities of intelligent agents". He was eventually fired.

That being said, I think that in general companies give short shrift to architecture. You can take a look at the end2end mailing list to see a lot of really bright people discussing fundamental architural issues of the Internet infrastructures. Without these people, the web wouldn't have gotten off the ground to begin with.

- Tal Software Architect Big American Company

-- Anonymous, April 18, 2001

[Paul] Ease up, folks. I'm pretty sure you're not seeing these "high" ideas on TV or in the newspaper.

---

[lusenet@chalcedony.com] Here's an example. I think the NY Times counts as a non-tech publication; don't you? And SOAP and XML-RPC are tools, not end projects.

The link didn't work (I'm not a NYT subscriber, so I get a login screen). I will point out, however, that it's obviously in the technology section, which isn't restricted to consumer electronics.

[lusenet@chalcedony.com] And I've seen plenty of non-tech industry writings about .NET and Hailstorm, that's for sure.

By all means, cite them. I'd be interested in seeing counters to my initial claim. The only stuff I remember seeing in print about either of these is either some article in _Wired_, or covers some aspect of them that relates more directly to everyday life (a court hearing, perhaps, or a lengthy explanation of such-and-such's impact on home life).

-- Anonymous, April 18, 2001

Peer to peer is not about creating revolutionary new services, it's about allowing currently existing services to survive. When Napster gets sued, a peer-to-peer system can replace it and provide the same functionality, but without a central server to sue. When Slashdot runs out of money, a peer-to-peer system can replace it, which is much more cost-effective and does not require big expensive databases and lots of bandwidth.

Peer to peer is not a hammer with new features; it's a hammer that provides the same old functionality but is made out of something much more durable.

-- Anonymous, April 18, 2001

[Paul] Ease up, folks. I'm pretty sure you're not seeing these "high" ideas on TV or in the newspaper.

[Dori] Here's an example. I think the NY Times counts as a non-tech publication; don't you? And SOAP and XML-RPC are tools, not end projects.

[Paul]The link didn't work (I'm not a NYT subscriber, so I get a login screen). I will point out, however, that it's obviously in the technology section, which isn't restricted to consumer electronics.

Jeez, Paul. You said that you didn't think we were seeing tools mentioned in the newspaper. I pointed to the NY Times. Now you're saying that because it's the technology section of the Times, it doesn't count as a newspaper???

[Dori] And I've seen plenty of non-tech industry writings about .NET and Hailstorm, that's for sure.

[Paul]By all means, cite them. I'd be interested in seeing counters to my initial claim. The only stuff I remember seeing in print about either of these is either some article in _Wired_, or covers some aspect of them that relates more directly to everyday life (a court hearing, perhaps, or a lengthy explanation of such-and-such's impact on home life).

Here's a Washington Post article. Yes, once again it's in the tech section, but that's a given, I think. How about a Business Week article? I've also found this Wall Street Journal article (via MSNBC) and this piece from Forbes.

BTW, I found these by going to NewsLinx and doing a search. Yeah, there were lots of techie publications, too, but there's no shortage of non-tech pubs talking about dev tools.

-- Anonymous, April 18, 2001

[Dori]Jeez, Paul. You said that you didn't think we were seeing tools mentioned in the newspaper. I pointed to the NY Times. Now you're saying that because it's the technology section of the Times, it doesn't count as a newspaper???

I'm sorry if I didn't make my point clearly enough. Here it is again: these abstractions tend to go only to the audiences interested in them. I cited trade rags, and should've included the tech sections of newspapers.

Which dodges the issue, sort of; you're likely a technical person (since you're reading this website), and Hailstorm doesn't interest you. But should it? I took the first (well, the second) link you graciously provided, to the Washington Post. Once there, I asked myself whether I, as a general technie who couldn't care less about the latest login protocol, would care at all about Hailstorm.

The truth is, I might, and the article says how I would. For anyone just getting in, Hailstorm is Microsoft's answer to the problem of universal identification on the Internet. (And so far as I can tell, they really DO mean universal this time, as opposed to "works with all Microsoft products", though we shall see how that plays out.) By Spolsky's metric, Hailstorm really does mean something: it means I can log into www.turbotax.com and see the status of my tax return without having to remember my ID and password that I had to make up specifically for TurboTax, and then later mosey on over to the Aussie Goat Cheese Home Page and see whether my shipment left the warehouse, again without needing the ID I made up over there.

Furthermore, there's implications on how payment-over-the-web is done, privacy issues, etc. Pretty relevant stuff if you ask me; in fact, it's relevant not just to techies, but anyone who does any transacting on the web.

As far as I can tell, Joel's beef with Hailstorm and other similar developments is that Average Joe can't do anything new right now just because Microsoft announced Hailstorm. Well, that's true, to an extent. Joe can't do anything unless Joe happens to be a business manager deciding what ID scheme to recommend to his web portal department, or a developer deciding roughly the same. BUT, Joe might care if Joe is worried about privacy issues. That's of general interest, and in my opinion is justification for that article.

As for exactly how Hailstorm does what it does from a technical standpoint - that's definitely one for the propeller-heads, and aptly enough, there's no mention of that in the Post article. (Personally, I'm inclined to believe it has zero technical cleverness; it's just another attempted common API, basically, with good old-fashioned software behind it to make it go.)

I could go into the other articles and see if they also fit my hypothesis, but I'm outta time for now.

[Dori] [good links snipped] BTW, I found these by going to NewsLinx and doing a search. Yeah, there were lots of techie publications, too, but there's no shortage of non-tech pubs talking about dev tools.

One for the bookmark list. :-) Thanks...

-- Anonymous, April 18, 2001

i admit i was once an 'architecture astronaut'. it happened for 2 years when i was doing software framework research work in an ex- company.

shoot me if you want. i deserved it.

anyway, i was thinking about a visual graph-based programming approach using nodes and connections. great way to 'see the whole picture', rather than reading one-dimensional text code. soon enough, i found out that stuff like IBM DataExplorer, Visualisation Toolkit uses something like that. Now stuff like Maya, Softimage uses it too. so i wasnt so original, but at least im glad its an idea that works. however, the point is, unless u apply an architecture in a practical, specific way, its as useless as it is cool.

-- Anonymous, April 19, 2001

I believe another element in architecture astronautics is religion.

I recently attempted to consult for a company that was trying to do the classic web-app interface to a CICS back end. Not exactly rocket science.

But the technical architect was so caught up in using J2EE technologies (when all they really had to make was a simple client/server application) that it was not possible for anyone to see the forest for the trees.

In this particular case, I think the so-called "architect" was a shyster and was using obtuse and uninformed smoke blown out his ass to cover for the fact that he was useless :-)

-- Anonymous, April 19, 2001

Indeed. Thank you, Joel. And here I was wondering if I was thick for failing to keep up with the ever shifting blizzard of Java "standards", failing to understand what real benefits SOAP brings over XML-RPC, or thinking XSLT is a bag of shite.

-- Anonymous, April 19, 2001

Gee, I wonder why Dave Winer didn't plug this article on Scripting.com!!!

-- Anonymous, April 20, 2001

Bang on! Peer to Peer is the *last* thing Napster is about. How many people realise, that most users couldnt care a hoot about where those MP3's come from? I doubt if even Napster realises it. All they are doing is "downloading" songs.

IMHO, architectures that set out to solve more than one simple problem fail and flop miserably. There's an enormous benefit in barely planned systems that evolve over a period of time (like TCP-IP, which really is a cludgy protocol, but is supremely elegant in its interoperability).

Also, no-one seems to be studying the effect of prolonged exposure to hype. I'm quite surprised and bewildered, by end-user's insistence on Intel processors, for example. With each generation being more technologically aware, shouldnt someone study the influence of hype on technology adoption ?

-- Anonymous, May 12, 2001