It’s my contention that if a technology deserves market success, it has a good chance of getting it. Eric seems to find this counter-intuitive, or at least contrary to experience. So let me clarify by saying that this only applies to technologies and products that are or can straightforwardly be made to be SUFFICIENTLY COMPLETE, and that I mean this in the Geoffrey Moore/Crossing the Chasm sense of “whole product”. (Yeah, yeah, I know Moore didn’t innovate the “whole product” idea — but unless you recall who actually DID, please don’t object to my giving him credit for interpreting and popularizing it.)

No, no. I appreciate it. Sometimes I need a good thwack with the cluebat.

And I really appreciate your main point. While I don’t expect the folks who’ve been touting RT to publicly eat crow, I hope they’ll at least refrain from making new spoutings on the subject.

As for Fabian feeling embarrassed — I guess for that to happen he’d first have to believe that any of his failures or errors were, at least partially, his fault. And not just in the sense that he didn’t realize how awful people were, or the industry was, or universities were, or the United States was, or whatever …

CAM

Embarrassing you is the last thing I meant to do. Embarrassing Fabian would be more interesting, but would be much harder, for obvious reasons in no way related to his overarching intelligence.

While I’ve been saying “Columnar systems have their advantages in query performance, and RT’s seems to have been a clever columnar architecture,” I didn’t focus on the point that one of those advantages is the ability to make more of the reads sequential.

Considering the fuss I’ve been making recently about sequential-vs.-random, that’s pretty embarrassing.

CAM

Your opinion is of course shared by most former company insiders. I strongly
suspect it is the correct one, although I don’t want to say so emphatically
because I’ve really only heard one side of the story.

Of course, that’s not due to lack of trying. I attempted to contact the
remaining shell of the company — i.e., Steve Tarin — for comment,
and never heard back. And I repeatedly asked Fabian Pascal to give some kind
of evidence for his claims; that naturally got me nowhere.

Where I get confrontational about TransRelational is when people like Date
or Pascal claim that it is relevant to a critique of OLTP DBMS products, and
especially when Date charges sums of money for presenting this idea that only
makes sense to the payers if they’re learning something that will help them
manage their businesses in the near future. That’s just a wrongful act on
his part, and I hope that as more light as been shed on his actions, he’s
cut it out for good — unless, of course, somebody DOES advance the technology
some day. I’m not aware of him selling any full-day seminar tickets about
TransRelational since last fall, so maybe we’ve made progress on that front.

There are a LOT of different DBMS architectures that get significant amounts of vendor and/or investor cash. The reason they get this cash is because people who are (at least as a group) well-qualified to make such judgments believe they can be successfully and profitably sold.

Probably true, but I’m not disputing this, except perhaps the definition of “such judgments” (judgments about what? I was referring to the apparently unconscious shift from discussions about technical judgments, into those about market judgments).
Presumably vendors and investors are doing some technical analysis to determine feasibility, but profitability depends also (primarily?) on the perceived state of the market, the perceived ability to spur demand, etc. I’d wager that the final decision has lamentably little to do with the technical quality of the product or its ideas. Saleability is key, right?

“True relational” and/or “TTM” aren’t just losing out to one or a few quasi-relational approaches; they’re losing out to DOZENS of other technical strategies.

Losing out? SQL is still top dog, and while I have little evidence other than its history (displacing other “data models” in short order), I believe it’s based on its proximity to relational, rather than its distance. You know the DBMS market far better than I do, and I’ve seen your discussions of XML and OLAP and other features, but by and large they’re built atop SQL DBMSs. It’s difficult to have the discussion without separating add-on tools, fundamentally different data models, and implementation details - and the way they’re bundled.

So what are the criteria by which one judges whether a data management product can be successfully and profitably sold? Those criteria (and here I’m speaking from a great deal of personal expertise, as is evidenced by my resume of consulting to the industry) really do revolve around whether or not it would be economically benefecial to enterprises to use these products.

Not directly; or at least it’s an assumption. Why do you say they “really do”? How do you know, or how would you know otherwise if that were the case? Selling a product doesn’t require any such knowledge - or at the very least, they require only a vague belief in the efficacy of the product. They also don’t require any comparison; all a sales force has to believe is that they want to sell. Hopefully they believe in the product, or at least their company; I’m not positing widespread deceipt here. But selling people what they want, however noble, has no bearing on whether they actually need it, nor on whether better solutions exist, nor whether the salesman knows anything other than what they’re trying to sell.
Once more, with feeling: I’m not painting salesmen and consultants as snake oil salesmen here. I’m simply pointing out that marketing and hype exert a strong influence, and if your audience seems to want X(ML), why should you deny them? Why deny them what they want, when it’s so obvious? Why go through the pain of a careful comparison to alternatives? Why gamble the company’s investments on new technologies not related to past successes? There are perfectly rational (though disappointing) reasons to do the obvious and uninspired thing. It’s the role of consumers to educate themselves (and educators to inspire a love of learning), though we’re often disappointed in the results. I expect companies, and the market, to do exactly what you’re describing! I’m only assuming the most beneficial of motives to such a position.

Now, relational advocates draw a distinction that I think can be fairly accurately paraphrased as distinguishing between “Usefulness in the real world” (what The Market is an almost-undeniably good measure of)

This is just silly. Every new technology (and I’m not referring to simple one-offs on existing stuff) begins with a period of high risk, low acceptance, and perhaps the complete absence of a market. How would you have ranked the “usefulness” of computers, or ARPANET, or early stem cell research, using commercial knowledge? Again, “good measure” as compared to what? At the very least you have to admit that the most popular products generate press and copycat products, leading to a dilution in available media of information about altneratives. Clearly if a product stinks, consumers will abandon it; the ones that stick and stink are those that are just barely good enough, and better than what came before. I think you underestimate the power of “just barely good enough.” The world will beat a path to your door if you have a better mousetrap, but when transition costs are high, only if their current mousetrap really sucks.
And even assuming you know the market can properly measure “usefulness” (I’m curious how you know it’s a good measure, rather than a poor or mediocre one, and what you’re comparing it with), over what time frame? Has relational had enough time? Too much? Are good ideas ever resurrected decades after their inception and initial failure?

and “Usefulness in The World The Way It Should Be”). But before I go on to address that counterargument, I’ll pause to check whether it’s the counterargument you’re actually making.

No, I’m not going to make crank rants about the way the world ought to be, though each of us probably has our thoughts on that subject. I’m not even railing against the market; simply disputing its value as a fetish for either group of extremists, and advocating that we, as technologists in various degrees, at least try to advocate technologies on technical merits, rather than blindly following the market. Yes, we all have to eat, but it’s undeniable that passion and belief and dedication (however dryly expressed by geeks and nerds) have had some role in technological development. Without it, there’s nothing to market. It just appears to be that the balance of attention even in the development community has shifted unnecessarily away from technical merit toward profitability, and thus hype and groupthink thrive.

While my views are similar to Scott’s, they’re not identical, and I’d rather defend my version of the ideas than his.

But I do think you’re underrating what can be inferred from the lack of vendor adoption of certain technical strategies. There are a LOT of different DBMS architectures that get significant amounts of vendor and/or investor cash. The reason they get this cash is because people who are (at least as a group) well-qualified to make such judgments believe they can be successfully and profitably sold. “True relational” and/or “TTM” aren’t just losing out to one or a few quasi-relational approaches; they’re losing out to DOZENS of other technical strategies.

So what are the criteria by which one judges whether a data management product can be successfully and profitably sold? Those criteria (and here I’m speaking from a great deal of personal expertise, as is evidenced by my resume of consulting to the industry) really do revolve around whether or not it would be economically benefecial to enterprises to use these products.

Now, relational advocates draw a distinction that I think can be fairly accurately paraphrased as distinguishing between “Usefulness in the real world” (what The Market is an almost-undeniably good measure of) and “Usefulness in The World The Way It Should Be”. But before I go on to address that counterargument, I’ll pause to check whether it’s the counterargument you’re actually making.

My mistkae

Um… yeah. These conversations with myself are apparently degrading to both of us.