## Rounding and heaping in non-software estimates

Round numbers are often preferred in software task estimation times, e.g., 1, 5, 7 (hours in one working day), and 14. This human preference for round numbers is not specific to software, or to estimating. Round numbers can act as goals, as clustering points, may be used more often as uncertainty increases, or be the result of satisficing, etc.

Rounding can occur in response to any question involving a numeric value, e.g., a government census or survey asking citizens about their financial situation or health. Rounding introduces error in the analysis of data. The Whipple index, described in 1919, was the first attempt to quantify the amount of error; calculated as: “per cent which the number reported as multiples of 5 forms of one-fifth of the total number between ages 23 to 62 years inclusive.” for errors of reported age. Other metrics for this error have been proposed, and packages to calculate them are available.

At some point (the evidence suggests a 1940 paper) a published paper introduced the term heaping effect. These days, heaping is more often used to name the process, compared to rounding, e.g., heaping of values; ‘heaping’ papers do use the term rounding, but I have not seen ’rounding’ papers use heaping.

The choice of rounding values depends on the unit of measurement. For instance, reported travel arrival/departure times are rounded to intervals of 5, 14, 30 and 60 minutes; based on reported/actual travel times it is possible to estimate the probability that particular rounding intervals have been used.

The Whipple index fails when all the values are large (e.g., multiple thousands), or take a small range of values (e.g., between one and twenty).

One technique for handling rounding of large values is to define roundedness in terms of the fraction of value digits that are trailing zeroes. The plot below shows the number of households having a given estimated balance on their first mortgage in the 2013 Survey of Consumer Finances (in red), and the distribution of actual balances reported by the New York Federal Reserve (in blue/green; data extracted from plot in a paper and scaled to equalize total mortgage values; code+data):

The relatively high number of distinct round numbers swamps any underlying distribution of actual values. While some values having some degree of roundness occur more often than non-round values, they still appear less often than expected by the known distribution. It is possible that homeowners have mortgages at round values because they of banking limits, or reasons other than rounding when answering the survey.

The plot below shows the number of people reporting having a given number of friends, plus number of cigarettes smoked per day, from the 2015 survey of Objective and Subjective Quality of Life in Poland (code+data):

The narrow range of a person’s number of friends prevents the Whipple index from effectively detecting rounding/heaping.

The dominance of round numbers in the cigarettes smoked per day may be caused by the number of cigarettes contained in a packet, i.e., people may be accurately reporting that they smoke the contents of a packet, rather than estimating a rounded number.

Simple techniques are available for correcting the mean/variance when values are always rounded to specified boundaries. When the probability of rounding is not 100%, the calculation is more complicated.

Rounded/Heaped data contains multiple distributions, i.e., the non-rounded values and the rounded values; various mixture models have been proposed to fit such data. Alternatively, the data can be ‘deheaped’, and various deheaping techniques have been proposed.

Given the prevalence of significant amounts of rounding/heaping, it’s surprising how few people know about it.

## Visual Lint 8.0.9.353 has been released

Visual Lint 8.0.9.353 has now been released.

This is a maintenance update for Visual Lint 8.0, and includes the following changes:

• Added support for Codegear C++ Builder 11 Alexandria.

• Fixed a crash in the Visual Studio plugin when a file was opened in Visual Studio 2022 v17.2.0 Preview. The crash was apparently caused by a change or regression in the VS2022 implementation of `EnvDTE::DocumentEvents::DocumentOpening()`.

• When analysing Visual Studio projects the appropriate `-std` and `-m32` or `-m64` switches are now automatically added to generated Clang-Tidy command lines.

• Fixed a bug in the Clang-Tidy analysis results parser which caused issues containing multiple square brackets to be parsed incorrectly.

• Updated the PC-lint Plus compiler indirect file `co-rb-vs2022.lnt` to support Visual Studio 2022 v17.1.6.

• Updated the PC-lint Plus compiler indirect file `co-rb-vs2019.lnt` to support Visual Studio 2019 v16.11.13.

• Updated the PC-lint Plus compiler indirect file `co-rb-vs2017.lnt` to support Visual Studio 2017 v15.9.47.

• Updated the PC-lint Plus compiler indirect files `co-rb-vs2017.lnt`, `co-rb-vs2019.lnt` and `co-rb-vs2022.lnt` to limit the value of `_MSVC_LANG` to `201703L`. This is necessary as the version of the Clang front-end used by PC-lint Plus 1.4.x and earlier does not include C++ 20 compiler intrinsics.

• Updated the installer to terminate any instances of `mspdbsrv.exe` (the Visual Studio program database symbol server) before installing the Visual Studio plugin to VS2017/19/22.

• Updated the text on the "Analysis Tools" page of the installer.

• Updated the readme and help documentation to reflect the fact that Windows 11 is now supported.

## Complex software makes economic sense

Economic incentives motivate complexity as the common case for software systems.

When building or maintaining existing software, often the quickest/cheapest approach is to focus on the features/functionality being added, ignoring the existing code as much as possible. Yes, the new code may have some impact on the behavior of the existing code, and as new features/functionality are added it becomes harder and harder to predict the impact of the new code on the behavior of the existing code; in particular, is the existing behavior unchanged.

Software is said to have an attribute known as complexity; what is complexity? Many definitions have been proposed, and it’s not unusual for people to use multiple definitions in a discussion. The widely used measures of software complexity all involve counting various attributes of the source code contained within individual functions/methods (e.g., McCabe cyclomatic complexity, and Halstead); they are all highly correlated with lines of code. For the purpose of this post, the technical details of a definition are glossed over.

Complexity is often given as the reason that software is difficult to understand; difficult in the sense that lots of effort is required to figure out what is going on. Other causes of complexity, such as the domain problem being solved, or the design of the system, usually go unmentioned.

The fact that complexity, as a cause of requiring more effort to understand, has economic benefits is rarely mentioned, e.g., the effort needed to actively use a codebase is a barrier to entry which allows those already familiar with the code to charge higher prices or increases the demand for training courses.

One technique for reducing the complexity of a system is to redesign/rework its implementation, from a system/major component perspective; known as refactoring in the software world.

What benefit is expected to be obtained by investing in refactoring? The expected benefit of investing in redesign/rework is that a reduction in the complexity of a system will reduce the subsequent costs incurred, when adding new features/functionality.

What conditions need to be met to make it worthwhile making an investment, , to reduce the complexity, , of a software system?

Let’s assume that complexity increases the cost of adding a feature by some multiple (greater than one). The total cost of adding features is:

where: is the system complexity when feature is added, and is the cost of adding this feature if no complexity is present.

, , …

where: is the base complexity before adding any new features.

Let’s assume that an investment, , is made to reduce the complexity from (with ) to , where is the reduction in the complexity achieved. The minimum condition for this investment to be worthwhile is that:

or

where: is the total cost of adding new features to the source code after the investment, and is the total cost of adding the same new features to the source code as it existed immediately prior to the investment.

Resetting the feature count back to , we have:

and

and the above condition becomes:

The decision on whether to invest in refactoring boils down to estimating the reduction in complexity likely to be achieved (as measured by effort), and the expected cost of future additions to the system.

Software systems eventually stop being used. If it looks like the software will continue to be used for years to come (software that is actively used will have users who want new features), it may be cost-effective to refactor the code to returning it to a less complex state; rinse and repeat for as long as it appears cost-effective.

Investing in software that is unlikely to be modified again is a waste of money (unless the code is intended to be admired in a book or course notes).

## Chasm City

Alistair Reynolds
ISBN-13‏: ‎ 978-0575083158

Following the announcement of the release of Inhibitor Phase and then Elysium Fire I’ve been rereading some of the previous Revelation Space novels to pick up the thread. First time around I found Chasm City a dark story and it was no different the second time, but I got so much more out of it. I also remember losing the thread towards the end the first time, but not this time!

As with most of the series, the thread of the main story is inconsequential to the main Revelation Space arc. It’s the other aspects of the story which tie up with other Revelation Space events which make this such a fantastic book. By the time I read the last page I knew that Sky's Edge was named after the edge Sky Hassausman had over the other ships in the flotilla which settled the planet. I knew that the war had started between the ships of the flotilla and what they were fighting about. I knew how the Melding Plague had got to Chasm City and how it was discovered and spread. I knew that Sky had met Khouri, who is an important character in the main trilogy. And more, much more! I wish I knew how the Melding Plague came to be though.

I read the second half of the book in about two weeks. I just couldn’t put it down!

## Pathfinder – baron m.

Welcome Sir R-----! Pray shed your overcoat and come dry yourself by the fire. I am told that these spring showers are of inestimable benefit to farming folk, but I fail to grasp why they can’t show the good manners to desist until noblemen have made their way indoors.

Will you join me in a warming measure and perchance a small wager?

I’ve a mind for a game oft played by the tribesmen of Borneo upon the cobbled floors of their homes as a means of practice for their legendary talent in forging paths through the dense forests in which they dwell.

## The Excess Strategic WIP problem

Try pouring a bottle of milk into a glass with milk already in it. You have a choice: stop or tidy up the mess afterwards. That is my work-in-progress (WIP) analogy, if you try and do too much – no mater how much you want to do it – you will end up with a mess.

Agile folk are well versed in the problems created by too much WIP and how to deal with it – check out the Stockless Production video if you want to see. In the last six months I’ve been seeing a particular variant on this problem with I’ve come to label the Excess Strategic WIP problem.

In the latest report the manager told me how the team completed a great quarter with 3 priorities – set via OKRs. The senior management team were so impressed they asked for 19 priorities in the next period.

Right now I don’t have a “paint by numbers” solution, fixing this problem is more involved. I’m starting to understand it and I’ll make some suggestions later. Ultimately this a failure of leadership to say “No”. That failure is itself rooted in a failure of leadership to appreciate what is happening on the ground and what doing the worker are up against: the number of strategic initiatives or the amount of business as usual.

In a team it is easy to spot excess WIP using a visual system like a whiteboard or card system. When you track work like this you see an awful lot of work sitting in the “in progress” column. Typically there is more work than people on the team and the work isn’t moving. Some of the work may be actively marked as blocked but more likely most of it is nominally being worked on.

It can’t all be worked on at the same time because despite having two eyes, two hands, and two sides of a brain human’s can only really do one thing at a time – even new parents. I label this work “WHIP” – work hopefully in progress. While it is, in theory, being worked on, most of it is just sitting there waiting for a multi-tasking (i.e. time slicing) person to come back to it.

The good news is when you can see the WHIP you are half way to solving the problem: there are well known solutions. Accept less work, impose work in progress limits, sequence work by adding queues to the board, educate people to work on one thing until it is done, etc. Once you can see the work you have a feedback mechanism, you can take action and, thanks to the feedback mechanism, watch it reduce.

But Strategic WIP is more difficult. Strategic WIP is the stuff the organization decides in really important, the stuff the most senior leadership decides should be happening. The Excess Strategic WIP Problem occurs when those strategic priorities are greater than the organisation’s ability to deliver on them.

While some of this work may be transactional (“Build a Mega Widget”) much of it is transformational (“Adopt Agile working”, “Increase diversity”, “Tighten security”). Such things involve changing the way other work is done: changing the processes, changing the criteria, increasing awareness. The feedback cycle is long but to get started people need to devote time: attend training, arrange kick-off meetings, discuss approaches with consultants/coaches, etc.

In one case I saw this year the organization has, for years, been asked to do more than it is resourced to do. While a few months of such a mismatch might have been manageable the cumulative effect has been to create organizational debt and demoralised staff.

The second case I saw isn’t a result of under resourcing, if anything that organization has too much – money, people and perhaps equipment then they know what to do with. But because their management model resembles a sponge it is impossible to know when the organization has taken on too much. While some pockets were overworked I’m sure other pockets were idle.

In both these cases “lean” was a dirty word. Both organizations had been subject to lean programmes that had stripped out “waste” but, from what I could see, removing that waste had created both organizational debt and demoralised staff. Having removed “spare” staff those left were juggling. Staff didn’t have time for “agile” and their diaries had no space for daily essentials let alone another change programme.

The third case came to light when discussing OKRs. The company had a successful quarter were they focused on a few objectives that success seems to have bred the next failure when the organization requested many objectives.

Actually, this case brought it home: taking on too many, or being given too many, OKRs. I’ve heard of teams tackling too many OKRs so many times in the last year. (In fact, I should name this “Excess OKRs Problem”. This occurs whenever you have more OKRs than you can count on one hand. Don’t tell me your team is big enough to do so, check out Focus is not divisible so limit you OKRs.)

In a way, the Excess OKRs Problem is better than the Excess Strategic WIP Problem because you can see it. One can say “OK company, you have asked us to deliver 15 OKRs here, we need to sit down and talk about this.” Like visualising your work in progress excess OKRs is a thing you can identify and address, it is the reason to talk.

(Note: unlike User Stories which should always be small, OKRs should never be small. OKRs should be big, meaningful and preferably strategic. No team can take on 16 meaningful OKRs, even 5 is too many.)

One of the problems with excess Strategic WIP is that it can be difficult to see, there are different teams involved and in different places. Conflicting priorities are hidden. People on the ground may see problems but the senior people – the people creating the WIP – are too far removed. Those people may not want to hear people saying “You are asking too much”. They may have too much riding on getting multiple work streams done. They may be deaf to the cry of pain when people say “too much.” They may have too big an ego to accept that what they are asking for is a problem. And it may be politically unacceptable.

“Political” is an important word here: several of the cases I’ve heard of, and some of those example above, are Government agencies.

Because excess strategic WIP is difficult to see it is difficult to build a feedback loop and difficult to take action. How do you know when the problem is too much WHIP and when people are “crying wolf” ? – which in itself implies a problem of trust and maybe a belief that “everyone is lazy, we need to push harder.”

By its nature “strategy” is big, which means that the feedback cycles are long and the problems of excess strategic WIP take time to play out. What is a WIP problem looks like another failed strategy.

While I would like to think OKRs can help with this situation – because they force teams and organizations to take stock of what they are working on – they may be making things worse because OKRs include an ambition agenda. Teams are encouraged to “shoot for the moon” and build “10x solutions”. There is good logic here: if one aims for a “10x solution” (i.e. a solution 10 times better than the status quo) and falls short the “failure” may still be “better” (e.g. “5x”) than if one had aimed for a “2x solution” and succeeded.

Ambition with OKRs should not be about doing more OKRs, rather ambition is within the OKRs, a challenge that makes you approach it differently. One can draw a line here between having one OKR which aims for “10x” and having 10 OKRs but I suspect the subtlety will be lost on someone asking for “more than you think you can do.”

So whats is the solution?

I’m not sure there is a silver bullet but I would want to … make the problem visible, perhaps a portfolio level kanban board. I would want to build a feedback loop so I could measure change. I would want to show the strategic people the visualisation.

Management education has a part to play too. I can believe many senior managers would benefit from understand what WIP is, the problems excess WIP can cause, the way excess WIP plays out on a day-to-day basis and how it effects people’s working lives. And perhaps most of all, address trust and the belief that “we just need to push harder.”

That might also mean some of the Lean waste lessons and OKR ambition lessons need to be revisited.

The post The Excess Strategic WIP problem appeared first on Allan Kelly.

## A new career in software development: advice for non-youngsters

Lately I have been encountering non-young people looking to switch careers, into software development. My suggestions have centered around the ageism culture and how they can take advantage of fashions in software ecosystems to improve their job prospects.

I start by telling them the good news: the demand for software developers outstrips supply, followed by the bad news that software development culture is ageist.

One consequence of the preponderance of the young is that people are heavily influenced by fads and fashions, which come and go over less than a decade.

The perception of technology progresses through the stages of fashionable, established and legacy (management-speak for unfashionable).

Non-youngsters can leverage the influence of fashion’s impact on job applicants by focusing on what is unfashionable, the more unfashionable the less likely that youngsters will apply, e.g., maintaining Cobol and Fortran code (both seriously unfashionable).

The benefits of applying to work with unfashionable technology include more than a smaller job applicant pool:

• new technology (fashion is about the new) often experiences a period of rapid change, and keeping up with change requires time and effort. Does somebody with a family, or outside interests, really want to spend time keeping up with constant change at work? I suspect not,
• systems depending on unfashionable technology have been around long enough to prove their worth, the sunk cost has been paid, and they will continue to be used until something a lot more cost-effective turns up, i.e., there is more job security compared to systems based on fashionable technology that has yet to prove their worth.

There is lots of unfashionable software technology out there. Software can be considered unfashionable simply because of the language in which it is written; some of the more well known of such languages include: Fortran, Cobol, Pascal, and Basic (in a multitude of forms), with less well known languages including, MUMPS, and almost any mainframe related language.

Unless you want to be competing for a job with hordes of keen/cheaper youngsters, don’t touch Rust, Go, or anything being touted as the latest language.

Databases also have a fashion status. The unfashionable include: dBase, Clarion, and a whole host of 4GL systems.

Be careful with any database that is NoSQL related, it may be fashionable or an established product being marketed using the latest buzzwords.

Testing and QA have always been very unsexy areas to work in. These areas provide the opportunity for the mature applicants to shine by highlighting their stability and reliability; what company would want to entrust some young kid with deciding whether the software is ready to be released to paying customers?

More suggestions for non-young people looking to get into software development welcome.

## Return of the sprint goal? (Infographic)

Most of the sprint goals I’ve ever seen are rubbish. Pretty much “do the random collection of stuff we’ve decided already.” Such goals are meaningless – save time by skipping them.

If a team adopts OKRs then I really hope they move towards more goal based working, in which case the sprint goal starts to have meaning again – although maybe the OKR replaces the sprint goal?

Either way, I see an opportunity to move away from backlog driven development (BLDD) and towards a more purposeful style of working. So it was interesting a few weeks ago when Gareth Davies from Parabol (online meeting exercises and such) sent me over this infographic and a link to a blog on Sprint goals. Food for thought.

The post Return of the sprint goal? (Infographic) appeared first on Allan Kelly.

## How I write books (my new book)

Regular readers will know I write books – quite a few by now, it gets embarrassing.

Being an author is a great conversation starter, when people hear you’ve written a book or two they want to know more – everyone seems to have a dream of writing their own book. It also means that people seek me out to ask my advice about a book they are writing, or thinking about writing.

So, I’ve started to write down all the advice I give to people in a new book – How I write books.

I’m following my usual pattern so you can buy early versions on LeanPub now – I released it last week and it immediately sold a few copies. As usual at this stage everything is in a state of flux; spelling, punctuation, grammar and all that jazz will be fixed later. Of course, anyone buying now will get free updates as they become available all the way to the final version.

If you do buy, then please let me know what you think.