Freesteel Blog » My overwhelming feat of insignificance

My overwhelming feat of insignificance

Thursday, February 28th, 2013 at 7:01 pm Written by:

Work that is not productive, such as duplicating someone else’s work, subtracts from work that is productive. This matters when you are getting old and tired and no longer able to catch up on mistakes by programming all hours of the night.

I feel it is necessary to spend excessive time working out what needs to be done due to the dire shortcomings of the internal comms system. It’s easy to know where you are in a small company, because there are so few of you and the coverage is predictable. Duplication of other work (done by people in another company) takes place consciously and strategically. But in a large organization you can find 20 people smarter and younger than you each day, and there is no way that one of them hasn’t already programmed what you are about to waste the next two weeks programming. What is it again? Have I got a triangle colliding with a cone that is being swung on the end of a stick? I can easily enter a state of paralysis.

I continually look for clues everywhere. Last night I listened to the 4th quarter earnings report where the CEO read out his deadly dull prepared statements, and then got quizzed by the owners of the company.

Let’s take one of their questions:

Steven M. Ashley – Robert W. Baird & Co. Incorporated, Research Division – I’d just like to ask about the domestic commercial construction business. You’re probably aware, we’ve seen some encouraging data points from the ABI data. Just wondering what your take is on the health or probability of a — seeing a cyclical recovery domestically in commercial construction during calendar ’13?

Carl Bass – Chief Executive Officer, President and Director – Yes, Steve. I would say, we’re seeing the same thing that you are. Certainly, some of the indicators are up. We’re also anecdotally hearing that pipelines are growing. That hiring is starting. So I mean, we’re seeing good signs, and I tried to refer to those earlier in my remarks that both in manufacturing and in commercial construction, particularly in the U.S., we’re seeing good signs. I don’t think we’re out of the woods yet. The optimism I hear from our customers is just tempered somewhat by the stupidity I see by our politicians.

As I have said, these money men don’t know jack about how to value a company. Otherwise they would operate with numerical models that required numbers, and unexplained departures from those models would cause them to be junked and replaced by new ones after a period of fevered creativity — as happens in science or engineering. What is this garbage — “encouraging data points”, “good signs”, “out of the woods”? If you were talking about the health of a living being it would be phrased in terms of a body-fat index or a T-cell count.

The absolute values are wrong, and probably wrong all the time. But everyone in the market uses the same wrong numbers, so it doesn’t make a difference. They only operate on “getting better” or “getting worse” to decide if it goes up or down, like velocity but not position of a yo-yo. And for this, the comparison values, they wait to get given the sales targets for the year by the same company directors who are then held to account on those figures which they had chosen. The money men have done nothing, literally nothing for their pay. They are con-men, like those multiple chess players who copy the moves from one board to the next so that they are mathematically guaranteed to win half their games.

What about the operational effects inside a company, in terms of efficiency of software development? We don’t have metrics for that, so they get ignored. But it would be fun if they tried to work them out. For example, here is the infamous “WOW support plan” proposed by the part time patent troll Versata who somehow got the idea that software defects are created by users — rather than discovered like left-over turds under the pillow at great cost and inconvenience:

When a user reports bugs (or files tickets), or outlines feature requests, they are doing you a favour. They ought to get a discount, not be charged extra.

There is a qualitative transition point between bespoke software for a single business user who must pay the costs of development, and mass market software where potential customers are shopping around and can buy it if they like it. In the latter case you have got to get it right, because if you don’t, they won’t buy it, and they are under no obligation to tell you why. So when some of them take the time to tell what you don’t know about the defects they have noticed, you have got to pay attention.

Is the market competitive? Does the end user quality of the software matter? (It clearly does not for most Microsoft products.) Is there an open user forum that acts as a functioning feedback loop between the users who are generous with their feedback and the software developers who are the only ones who can do something about the defects? The owners could check this out. A high volume of extremely varied bugs being reported and dealt with satisfactorily is a good sign. That’s what effective work looks like.

Let’s take a bit more from the 4th quarter call-in, where someone is trying to decide whether the quoted sales growth of the company was due to its own products, or if it was actually coming from the sales of products made by companies that have been acquired throughout the year.

Richard H. Davis – Canaccord Genuity, Research Division – So like over the — it looks like over the last 3 quarters, you’ve made about $260 million, $263 million of acquisitions. So if I kind of do the math, I’m not — I don’t know how much you paid for all these companies, but some of the revenues would fall into last year, but some of it will fall into this year, of the 6% — 6% to 7% revenue growth, is it — am I close figuring maybe 1/4, so 2% or 1.5% of that is inorganic, so that the underlying organic growth is, I don’t know, 4% or 5% or something like that, is that a logical conclusion?

Carl Bass – Chief Executive Officer, President and Director – No, I would say, Richard, it’s closer to 1%. And given the number — I mean, we — so as we look at the year, there are a number of shifts going on in the business. Things that are coming out of the business. And on the plus side, I would say it’s not even quite 1%. Most of the acquisitions we tend to do are these small tuck-ins. We’re more positioning for the future. So when you look at it, for example, one of the acquisitions was the backbone for all of our social stuff. There were a couple of mobile acquisitions. There were a number in simulation that were really technologies. They’ll get folded in. That’s more typical of the way we do it. As you know, you’ve been following us, much more so in very few of the cases have we actually bought real revenue streams.

Davis – Got it. Okay. That makes sense. I just wasn’t sure what the ratios were, so that’s helpful.

Bass – It’s really relatively small and in many cases, it’s hard to parse even for us. For example, we make a Simulation 360 offering. It’s a combination of stuff that we’ve acquired over — acquired and built over a period of years. And truthfully, we don’t spend huge amounts of time trying to dissect it because it’s a fruitless exercise at some point. We try to look in aggregate at the benefit that we get from the investments in M&A.

Let’s have some fun by trying to work out where that 1/4 figure came from. It sounds like it’s back of the envelope type stuff. Here is what could be the source of his 6% figure from the CEO’s remarks:

The number comes from the table higher up the page (which he did not read aloud):

by virtue of the calculation (235-222)/222 = 5.8%. In other words, he cherry-picked the maintenance subscription revenue which went up, as opposed to the new sales revenue which stayed constant. It can be terribly easy to pump up the maintenance figures by some aggressive price hiking of your captured market, but this should by rights drive away potential new customers who have a choice not to have anything to do with that racket, so it is not necessarily a good sign.

But Davis isn’t making that call. Ignoring the fact that it appears he has mistaken this figure for the total revenue growth, he wants to know if some of the increase is actually a reflection of the sales/maintenance revenue made by the companies that have been acquired by Autodesk for $260million over the year.

So, if we begin with the basis that the $8billion market value of Autodesk yielded $235million of revenue, and we assume that $260million of acquired companies in the same business sector would yield proportionally 235*(260/8000) = $7.2million, then this would form the inorganic part of the $235million reported revenue. Therefore, of the (235-222)=$12million increase in revenue, more than half of it would be due to the absolute new revenue from the acquired companies. But since “some of the revenues would fall into last year, [and] some of it will fall into this year”, Davis halved the number and went with 1/4.

Now, you could have asked, “But what about the $221.7million of acquisitions that were disclosed in the January 2012 report?” Won’t those also fall into last year and the year before, so that the two halves from each year’s reporting would add together to make a whole, so that the number he should have prodded the CEO with was 1/2?

Evidently this data was not available to his scratch calculation due to a lack of structured data and computational models at a high level. For too long these finance people have expended humongous efforts programming computers to trade against one another on the scale of nanoseconds to the extent that they have never bothered making hay with the numbers disclosed in the real economy. Far too slow and boring, isn’t it? As I put forward in my ScraperWiki blogpost (that they were too chicken to publish for a month because they were frightened I would get the fired):

Basically, it’s clear to me that nobody in the primary finance system is using data or quantitative financial models to do their instant trading in response to the disclosures — or else these press releases and quarterly reports would look a whole lot different. They would be machine readable, wouldn’t they? With unique company number identifiers corresponding to structured assets of statistically approximated business models held by the investment houses that would enable the financial analysts to (a) instantly merge the known assets of the taken over company into the purchasing company, and (b) automatically value the sale-price assets of the taken over company in relation to its peers in the same market in the microseconds necessary to respond to the disclosure for the purpose of rapidly dumping or acquiring stock until the price is statistically close to the calculated fair value.

Yes, indeed, the identities of the companies which were taken over by Autodesk were disclosed at the time, but those geniuses at Cannacord Genuity and other money-heads don’t have the technology to deal with this information. Remember, “Ideas are the engine of our business” — because it sure ain’t the data. This isn’t a criticism of the analyst Richard Davis himself. The problem lies with the state of the art in the finance industry. Companies might as well be printing logarithm tables in their financial returns.

I mean, let’s just check out what revenue could be coming from this acquisition:

SAN FRANCISCO–(BUSINESS WIRE)–Oct. 4, 2012– Autodesk Inc., (NASDAQ: ADSK) has completed the acquisition of Qontext, enterprise social collaboration software, from India-based Pramati Technologies. The acquisition of the Qontext technology and development team will accelerate Autodesk’s ongoing move to the cloud and expansion of social capabilities in the Autodesk cloud-based service. Terms of the transaction were not disclosed

Autodesk intends to use the Qontext technology to add new social capabilities to Autodesk 360, a cloud-based platform that offers users the ability to store, search, and view critical design data improving the way they design, visualize, simulate and share work with others at anytime and from anywhere.

“Mobile, cloud and social computing are dramatically changing the way engineers, designers and architects work. The addition of the Qontext technology to the Autodesk portfolio will lead to new technology innovations that help our customers embrace these disruptive technologies and leverage them for competitive advantage,” said Amar Hanspal, Autodesk senior vice president of information modeling and platform products. “It was great to work with the team at Pramati who have demonstrated a great capability in incubating disruptive businesses.”

Actually, the terms were disclosed. For example, this is their website the day after:

So, not much revenue is to be expected in that sector from now on. And the Form 10-Q SEC filings disclosed:

On October 4, 2012, Autodesk acquired Qontext, an enterprise social collaboration software solution, from India-based Pramati Technologies for $26.0 million and hired the Qontext development team. This acquisition is expected to accelerate Autodesk’s ongoing move to the cloud and expansion of social capabilities in the Autodesk 360 cloud-based service. Treated as a business combination, Qontext has been integrated into, and the related goodwill was assigned to, Autodesk’s Platform Solutions and Emerging Business segment. The amount of goodwill that is expected to be deductible for tax purposes is $24.0 million.

According to Wikipedia:

Goodwill is an accounting concept meaning the value of an asset owned that is intangible but has a quantifiable “prudent value” in a business. For example, a reputation the firm enjoyed with its clients.

Now, if there is any goodwill reputation from closing down the product and practically dumping all your customers overnight, you’d think it would be minus several million, not plus 24. But this disreputable behaviour probably hasn’t caused the customers of the other Pramati portfolio businesses to look for the alternatives, because the information Is. Not. Joined. Up.

That’s why you don’t get laughed at when you write a press release that says information is not disclosed, when it actually is.

So what’s going on? Who can say? Let’s suppose the internal communication system within a large corporation is utterly diabolical beyond belief, as they normally are. Let’s suppose that in the middle of it is this god awful intranet system called SharePoint supplied by Microsoft that’s so bad it makes Windows 3.0 look like a marvellous piece of precision engineering. I don’t think this is a dangerous disclosure — after all, there are many companies even today who are just now installing SharePoint at the heart of their businesses. They’re like teenage school children coughing and spluttering on their very first cigarettes behind the bike shed, working hard to acquire a nasty habit that will damage them for the rest of their lives. It’s the way of the world.

Let’s suppose the company then buys an enterprise social collaboration software solution — including its entire development team, who are now no longer distracted by the product they used to be selling. Let us reasonably suppose the social collaboration tools they are working on are being trialled on the company staff. Let us imagine, as promised by the press release, that this development will “dramatically change the way engineers… work”. Could this be the most important acquisition of the year?

I’m not sure yet. Seems reasonable that technology intended to change the way your customers work really ought also to change the way you work, and the point of that is to improve productivity.

What was the Qontext product anyway? According to archive.org:

Stop digging your inbox
Each time you open Qontext Toolbox, you will see the most recent activity and any conversations or content that you may need while on a Salesforce page. Filtered and contextual information eliminates the need to dig through long-buried emails and file folders.
Seamless integration with Salesforce Objects
When your Salesforce Admin enables Qontext, you will notice that it appears seamlessly integrated as an open panel within your Salesforce objects such as Accounts, Leads, Contacts, Opportunities etc… This open panel, the ‘Qontext Inset’, is designed to appear as one of the sections inside the Salesforce object page. Qontext Inset displays real-time Salesforce object-specific content and activity.

Salesforce? What do we know about them, apart from their corporate logo looking like the rear end of a toy poodle.

Autodesk CFO Jumps to Salesforce.com (posted May 2002)
“Were going to miss Steve, but hes decided to move on to the startup world to join Salesforce.com as their chief financial officer,” said Carol Bartz, Autodesk chairman and CEO, in a statement. “We wish him the best, and as a Salesforce.com customer, we have a stake in his future success.”
Autodesk, also based in San Francisco, develops design software and digital content.

The pieces are beginning to come together. And they are not necessarily technology driven. But sometimes the pieces accidentally get put together — like about 10 years after they should have been, if folks were on the ball. And then one day there is a new internal comms system floating around that is being developed internally, so there is a chance it’ll be put right.

And then I’ll finally be able to find out what the heck is going on in here with all the other software engineers, so I don’t do something which someone else has done already last year. Yes? That’s what we all want.

I had an epiphany while I was writing this up this morning.

It was about Test Driven Development. We don’t want to get confused by the so-called rules that are unreasonably over the top. What is it really about? I think it’s about staking out a common domain in the hyper-dimensional space that is the software code space. You can plot the points for each repeatable automatic test onto a big sheet of 2-dimensional fabric. And these sample points can be a bit fuzzy if you include small random deviations in the variables of your tests, so you don’t always test the same exact calculations. Then, when you run your tests overnight, what you are really doing is intersecting this fabric at its sample points through that hyper-dimensional space against the functional manifold that is the current code state. And these points show up green, or red, or yellow if you need to take a closer look to check if the answer is right.

That flattened out test fabric. That’s the domain we work on. That’s the common map of the territory of where things are happening. The structure of the code, that changes quite a bit, if you are any good. If you are not doing much, then the code is not being refactored, and you can use the code as a fixed point of reference. In which case what exactly are you doing when you work on it day in and day out other than applying further patches onto a rotting superstructure. When you have more than two coders on the system, you need this domain fabric of tests, which stays fixed like a technical survey that we can all see and agree on. The design and the direction of the design can no longer exist in one guy’s head when more than one guy is working on it.

We got to have and present these maps and automatic verification suites up front. It’s like the foreman who compares the finished part against the blue-prints by measuring some dimensions with his callipers, not concerned about exactly how it got made. A big farm engines of tests and panels of outputs should be the first thing that is visible to any developer coming into the business, trying to find his way around, and looking for his niche in which to make a difference.

Sometimes open source projects will publish this common domain fabric, as VTK do.

A company that delivers a cloud services will normally disclose a small portion of the real time sample, as with status.salesforce.com and status.aws.amazon.com. (I have not been able to find the status page for Autodesk’s cloud services yet.)

We’ve got instances of this domain test fabric verifying the HSMWorks features, but I’ve never taken them very seriously. Now at last I know what they mean. And my period of dithering and indecision on the direction I have got to go in is coming to a close.

2 Comments

  • 1. Freesteel&hellip replies at 20th March 2013, 12:37 pm :

    […] actually gets disclosed, the stock market price doesn’t respond. The stock market analysts don’t care. So it’s a lie. The story is just plausible enough to keep the quivering animal ensnared in […]

  • 2. Freesteel&hellip replies at 30th April 2013, 10:52 pm :

    […] is limited due to effort being expended on an internal blog on the Qontext system in an attempt to annoy people who don’t know how to make a good interactive web […]

Leave a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <blockquote cite=""> <code> <em> <strong>