How to prove your programming experience?

[Guennadiy VANIN]

I developed (as well as maintained, debugged, used millions lines of) components for very big distributed on-line applications.
So big that nobody cared about quality of what is available for years. The work was in outsource and the stuff was very volatile either.

A few clicks in Visual Editor or wizard of JBuilder and you end-up with thousands lines of code for GUI, tables, arrays, forms, letters, frames ... – for printer, for online, for another application layers, etc.., etc.
The potential employer insists that in order to verify your competence you should submit your code samples.
Is it normal?
How it is possible, if you worked with millions lines of code?
How to prove your programming experience to others?

So, it never came to my head to bother about preserving in my disposal any codes or bringing it outside of office. Then I worked in outsourcing in installations/premises of client (really state Ministry of Economy, Inst. of Support of Small and Medium Enterprises and of Investments).
Well, I tried to respond that it is not my property and gave Internet links to front-end interfaces.
I also gave link to open source classes at my previous employer's site (not exactly mine, I have never bothered to fill my name as an author, and, I am afraid I could not since it is a matter of property of my employer).
Then who is the author of the code with thousands of changes by many programmers, with changes bigger than the final file.
I tried telling that I am SCPJP2. But I was told that they have 50 candidates, all claiming certified with lot of experience and it is necessary to validate candidature by submitting my samples for scrutiny.
I had never encountered such requests before ...
So, I never bothered to identify MY code, then even if I have bothered before, how can I present millions of lines?
The potential employer still insisted. And he asked me that it is impossible to verify veracity of my CV claims and capabilities.
I have in my CV links to my previous employer (and there any contacts).
Earlier with a lot of job offer, employers did not bother with all these and I hadn't headaches because there was plenty of other opportunities.
Now everything has changed. So, how should I present the codes?
Is it a matter-of-fact requirements?
How to prove that experience?
I have strange feeling that I miss something.
Is it reasonable to comply to such requests?

[Jon McDonald]

Actually I wish more employers would ask for sample code.
All you need to do is build an application using some of the technologies that you claimed expertise in on your CV. It doesn't have to be huge, but the application should not be too simplistic (no "hello world" programs).
Now that I think of it, it would probably be a good idea for any programmer to have a CD with sample applications dealing with all (or most) of the technologies they are skilled at (excluding server specific skills).
My advice would be to spend a weekend building an application which uses the skills that the employer is looking for (excluding probably application server specific items, like Weblogic stuff, unless you have a Weblogic app server handy .
In fact, I think I'll start building a CD like this for myself it will definately separate whoever has a CD like this from the rest of the applicants.

[James Hobson]

Of course most contracts will make sure that any code you write whilst employed is the property of your employer, and so it is likely that you are not allowed to show your work to anyone else.
Asking to see sample code isnt that helpful, unless you have time to ask complex questions about the code in interview.
I know people who have quite happily used code by someone else to show in interviews.

[Thomas Paul]

Show him a copy of Tomcat and see if he figures it out.

[Aruna Raghavan]

I had to take java tests twice last year during interviews. I ended up taking one of the offers and quitting after a year because it was a zombie company. I didn't mind it but there is something wrong if the company that wants to hires you does not know how to make sure to ask the right questions...

[Jon McDonald]

I really don't see the point in giving someone an additional test for a job interview when they have already taken the SCJP2. Especially since most of these tests have the same, computer based format.
It makes a lot more sense to have someone offer code samples and then ask them questions about the design decisions they made. While someone could try passing off someone elses code as their own, asking a few questions about their design decisions should weed out a lot of the cheaters.
There is also the idea of having someone write code during the interview. It would have to be simple to save for time. At first I thought that this could only be done in the final interview stages, but if employers have time to give people 3 hour computer based tests they have the time to give them three hour programming assignments. An automated testing framework could be used to check for validity of the program. If the person is taking the test at home they would be allowed to use whatever texts they want. If they are taking the test on site, javaDoc for the Java 2 API would be provided.
Now that is an idea, and it beats a multiple choice test.
Jon

[Michael Bronshteyn]

I am not sure that 3 hours test is something which will show developer's capability. In real programming world, where developer has to know OS ( UNIX, Windows ), database, servlet container ( or application server ), programming language ( Java, sometime C or C++ along the way ), design patterns, UML, database modelling one programming assignment is not enough.
On the other hand, if employer is prepared for interview, it is possible to ask questions about each topic and get a good picture on who is in front of you.
Also, besides programming skills, developer has to be a team player, and there is noway you can check it by giving programming assignment. In other words, no matter how much you know, if you are a pain in the neck, not many people would want to do deal with you anyway.

[Mark Herschberg]

This is common and good.
An engineer should be able to demonstrate some code s/he has written. Given that maintainability is the major cost in software, creating organized code is critical. Just as important, the engineer should be able to explain it.
Most companies would not be happy if you displayed their code. On the other hand, it's somewhat acceptable in the industry to bring in a small number of files (e.g. 5-6 classes) and show them. Most companies look the other way.
I strongly recommend creating a portfolio on a CD. It should include working applications, documentation, and source code (and resume). This is a portfolio you can bring with you to interview. You can ever leave it with them--but in that case, make sure everything on there is something you're allowed to show other people.
I have an open source project I've created. I can demo it, and walk them through the organization and source code.
--Mark

[Mark Herschberg]

Originally posted by James Hobson:
Asking to see sample code isnt that helpful, unless you have time to ask complex questions about the code in interview.

I've found it extremely helpful. Yes, you need to aks them questions. I don't have time to analyze it myself. But what I can do in about 15 minutes of walking through it with them is the following:
1) See their thought process, and how they organize their projects
2) Get a feel for how well they know the language/technology/API
3) See how organized they keep their code (e.g. naming, comments, cleanliness)

Originally posted by Jon McDonald:
I really don't see the point in giving someone an additional test for a job interview when they have already taken the SCJP2. Especially since most of these tests have the same, computer based format.

I do. I've met countless SCJP's who aren't competant programmers. But I'm no big fan of the test. In terms of testing in general, I ask questions not covered on the SCJP--that's what most interviewers do. But I assume you meant a raw programming test. Yeah, I don't need to test if he remembers how a Vector works, but ultimately the SCJP doesn't cover building a working project. it only covers basic Java grammar.
Overall I'm not a big fan of writing code during interviews. It's time intensive and personally, I don't think highly of them. A good interviewer should be able to determine the skill set from communicating and/or looking at source code. Those who can't shouldn't be doing interviews.

--Mark
[ January 03, 2003: Message edited by: Mark Herschberg ]

[Rufus BugleWeed]

I've met countless SCJP's who aren't competant programmers.

Do you have really high standards?
How many have you met that had 5+ years full life cycle development?
Would you hire a guy with J2ME experience to work an EJB project because it's like Matloff says -
any competent veteran programmer can become productive in a new programming language in a couple of weeks on the job

[Mark Herschberg]

Originally posted by Rufus BugleWeed:

Do you have really high standards?

Yes I do. What makes a company successful and what makes a job enjoyable are working with people you like. I like working with smart people.
Also, given the spread of develop ability contrasted with the spread of developer salaries, you get a much better bang for your buck at the upper levels.

Originally posted by Rufus BugleWeed:

How many have you met that had 5+ years full life cycle development?

I don't think I met any. If someone has 5+ years full life cycle experience and is reasonably smart, I doubt I'd care whether they have a certification or not.

Originally posted by Rufus BugleWeed:

Would you hire a guy with J2ME experience to work an EJB project because it's like Matloff says -
any competent veteran programmer can become productive in a new programming language in a couple of weeks on the job

Yes. I agree 100%, and have said the same thing myself many times. Now the presumption, just so we're clear, is that I have a smart guy with J2ME experience, and mediocre guys with EJB experience. Obviously given two equally smart guys and one with releveant experience, I'd take that one.
A smart guy can pick up the basics of a language or technology withing weeks (40-60% capacity). Become decent within a month or two (80% capacity), and get pretty damn good inside of 6 month (100% capacity).
Do the math. So the ability curve of the good guy new to the technology looks roughly like the following:
Month 1: 0
Month 2: 50% X
Month 3: 60% X
Month 4: 80% X
Month 5: 80% X
Month 6: 100% X
So over 5 months, the contribution is 2.7X. Or a waste of 2.3X. To many managers, this seems like a big waste. But now let's look at the not so great guy who is familiar with the technology
Month 1: 20% Y (still has general startup costs)
Month 2: 90% Y
Month 3: 100% Y
Month 4: 100% Y
Month 5: 100% Y
Month 3: 100% Y
His contribution is 4.1Y. If X=Y, thats 1.4 months of additional productivity. If X > Y then withing some number of months, we see the first developer overtakes the second.
The key question is, what is the ratio between X and Y. Obviously that varies for any two people. Demarco and Lister have empirically demonstrated a ratio of 2:1 in ability between the upper and lower levels of developers (i.e., given 100 developers separated in two camps, those above average and those below, on average someone in the better camp will be twice as productive as someone in the worse camp).

Bottom line, intelligence far outweights experience in most cases. Unfortunately, many managers are under so much pressure that they see the startupcost of training someone in a new technology is too high, because they take a short term view. They suffer in the long run.

--Mark

[Manish Hatwalne]

Originally posted by Mark Herschberg:

Bottom line, intelligence far outweights experience in most cases. Unfortunately, many managers are under so much pressure that they see the startupcost of training someone in a new technology is too high, because they take a short term view. They suffer in the long run.
--Mark

Couldn't agree more!!!
My question is - How to deal with such managers suffering from myopia? How to convince them that you can pick up the technology, only if given a chance to prove it!
- Manish

[Manish Hatwalne]

I hope the sample code that is being discussed here simply refers to the the code samples of the technologies mentioned on your resume, and not the sample code from the real life projects that you have worked for other clients. Which employer would want the employees to take the code outside?
Secondly, if you have worked on various diverse technologies, it might not be possible for you to reproduce correct code sample after a long gap, especially if you are not expecting a question on it. I have done a WAP based project approx 2 years ago, but haven't touched WAP after that, it might not be possible for me today to write correct code, esp if I am given to understand that the requirement is of Java/Servlet technology.
- Manish

[Pradip Bhat]

Both ,proper intelligence and proper experience matter.If only ,intelligence matters,why not hire Phds from any branch?(assuming any Phd has sufficient intelligence? ).I think while hiring, manager should rate the person according to
1)Problem solving skills(algorithms/analysis)
2)Writing the pseudo code
3)Understanding of his/her previous projects.
3)Documentation
4)Software engg.principles.

[Mark Herschberg]

Originally posted by rahul rege:
Both ,proper intelligence and proper experience matter.If only ,intelligence matters,why not hire Phds from any branch?(assuming any Phd has sufficient intelligence?

Because
1) PhDs from other branches aren't fluent in software. As much as I bash undergraduate CS programs, they do provide a basic framework for software. A history PhD won't know what a hashtable is. That adds another 6 months of education of 0 productivity.
2) PhDs from other fields don't have the right way of thinking. Programmers think in a very specific constrained way about problems. Other branches of study promote different patterns of thinking.
Let's just focus on EE/CS PhDs
3) There simply aren't enough.
4) Many have expertise in specific areas and aren't interested in being a programmer. They are put to better use elsewhere
5) Why get a PhD if you're just going to be a programmer. A PhD doesn't make you smarter, it just makes you very knowledgable about a certain problem domain. PhDs expect to be compensated for this extra knowledge. If that knowledge isn't relevant to the project, the employer doesn't get that extra value returned.

--Mark

[Mark Herschberg]

Originally posted by Manish Hatwalne:

My question is - How to deal with such managers suffering from myopia? How to convince them that you can pick up the technology, only if given a chance to prove it!

Run.
Seriously. Try to explain this to your manager. Recruit other engineers who agree and have them try. If, after a few months, your manager won't listen, and continues to make short sighted mistakes, leave the company.
(And for those saying, "but that's not so easy in a down economy" I say, "it's not that hard for a good programmer". I put my money where my mouth is and quit my job in Nov 2001 because I didn't like my managers decisions.)
--Mark

[Guennadiy VANIN]

You all, guys, jumped to the train of CDs.
I forgot to tell that request (and message exchange) was from foreign employer (really a university department) to send samples by Email. I understood that you all proposed me DOS (denial-of-service) attack at its sever.
I also would like to share you with parts of my SW experience:

FORTRAN77, PL/1, Assembler, Pascal: numerical methods, should I buy a nuclear power station for demo?

databases at Air Control Center. To buy a couple of airports fro demo?

OBLOG TOOLS (Portuguese competitor of Rational Rose): UML-conformant creation and testing of UML-conformant models in OBLOG. The problem that nobody seems to have specs on this language and tools and this language had not been commercialized.

Mathematica2x, 3.0, Reduce 3.0: mathematical evaluations

COBOL/CICS/DB2: should I buy mainframe? As a mtter of fact, in Bank with separation of Tests, Development, Integration and Production

JAVA (backofice application): should I buy an Economy Ministry where I worked? I worked at one layer without knowing the origin and destination of data flows.

As a matter of fact I even could not clear out what "employer" wants from me because the communication was at the level of secretary/HR girl dealing with 50-200 applicants/position. In the second round of Email exchange a guy entered. HR boss? They just wanted codes, the codes and nothing else but the codes.
Poste by Aruna Raghavan:
there is something wrong if the company that wants to hires you does not know how to make sure to ask the right questions...
Me too. But the time of wrongness is very wrong. It is unhiddenly wild capitalism currently out of my house. I even could not imagine a lot of requirements a year ago.
I wrote the same opinion that I do not want to work for the guys who discriminate candidates by the codes submitted through Email. As a matter of fact, I did not want to go to the more cold country with the winter coming.
As a matter of fact, I am so experienced that employers are always wrong and I am right... but out. They are wrong but in...

Posted by Jon McDonald:
Especially since most of these tests have the same, computer based format.

Is it in the same format that most exams are taken by all students?

Posted by Jon McDonald:
While someone could try passing off someone elses code as their own, asking a few questions about their design decisions should weed out a lot of the cheaters.

This is not correct. In production environment what counts is the functioning, the result but not why and what is the theory behind it. Even Senior Developers couldn't explain me why but just did know from predecessors that for functioning that and such steps should be made. Take it or leave it (i.e. you will be left)
This very strange irony of working in out-sourcing: I almost never learned anything from the work (if I did not know it already before)

Posted by Jon McDonald:
On the other hand, it's somewhat acceptable in the industry to bring in a small number of files (e.g. 5-6 classes) and show them.

Is it the company's (bad architectural) traditions or me who is evaluated? I was forced to work with codes written in the way of examples of how it should not been done (thousands lines per class, etc)
Then do not forget that in 95% of cases you do not speak with reasonable/intelligent people but with HR girls or even security guys who just distributes the forms for filling out.

Posted by Mark Herschberg:
Those who can't shouldn't be doing interview.

Do you mind if I refer to your name after using this phrase?

Posted by Mark Herschberg:
I like working with smart people.

I am anxious to know how do you distinguish the smart from just educated?

Posted by Manish Hatwalne:
My question is - How to deal with such managers suffering from myopia?

To be born rich and to buy his company. All the rest is just insignificant variations to correct result.
Mostly I worked in outsourcing:1-9 months here with these tools, 1-3 there, with those. Frequently I even could not get with what I have been working with....
[ January 03, 2003: Message edited by: G Vanin ]

[Mark Herschberg]

I've got to say, I don't understand half of what you wrote.
In any case, if you want to quote me, feel free, but you need to quote me sans grammatical error (even though it was my error), "Those who can't shouldn't be doing interviews."
As for distinguishing smart from educated, let's just say that those who are smart, are smart enough to know the difference.
--Mark

[Abadula Joshi]

Originally posted by Mark Herschberg:

Yes. I agree 100%, and have said the same thing myself many times. Now the presumption, just so we're clear, is that I have a smart guy with J2ME experience, and mediocre guys with EJB experience. Obviously given two equally smart guys and one with releveant experience, I'd take that one.
A smart guy can pick up the basics of a language or technology withing weeks (40-60% capacity). Become decent within a month or two (80% capacity), and get pretty damn good inside of 6 month (100% capacity).
Do the math. So the ability curve of the good guy new to the technology looks roughly like the following:
Month 1: 0
Month 2: 50% X
Month 3: 60% X
Month 4: 80% X
Month 5: 80% X
Month 6: 100% X
So over 5 months, the contribution is 2.7X. Or a waste of 2.3X. To many managers, this seems like a big waste. But now let's look at the not so great guy who is familiar with the technology
Month 1: 20% Y (still has general startup costs)
Month 2: 90% Y
Month 3: 100% Y
Month 4: 100% Y
Month 5: 100% Y
Month 3: 100% Y
His contribution is 4.1Y. If X=Y, thats 1.4 months of additional productivity. If X > Y then withing some number of months, we see the first developer overtakes the second.
The key question is, what is the ratio between X and Y. Obviously that varies for any two people. Demarco and Lister have empirically demonstrated a ratio of 2:1 in ability between the upper and lower levels of developers (i.e., given 100 developers separated in two camps, those above average and those below, on average someone in the better camp will be twice as productive as someone in the worse camp).

Bottom line, intelligence far outweights experience in most cases. Unfortunately, many managers are under so much pressure that they see the startupcost of training someone in a new technology is too high, because they take a short term view. They suffer in the long run.

--Mark

Excellent points !!! I wish all the employers can understand this.

[Rufus BugleWeed]

You can't measure the productivity of a developer. These are fractions of a mythical man month.

[Mark Herschberg]

Originally posted by Rufus BugleWeed:
You can't measure the productivity of a developer. These are fractions of a mythical man month.

Bullshit. You can measure developer productivity. you can debate about the accuracy of the measurements or its revelance to future performance, but you can most certainly measure it and in the extreme cases the difference is clear.
--Mark

[Rufus BugleWeed]

What are the units?
If a guy delivers a solution faster but has more rework and maintenance associated with it was he really more productive?
If a person does something creative, they can eliminate countless hours of the old way.
The story I heard goes something like this:
Henry Ford hired himself a management consultant. He and the consultant did some management by walking around. Everytime they went past a certain employees desk the employee had his feet up on the desk reading the newspaper. Consultant says to Henry, why don't you get rid of that guy he's not doing anything for you? Henry replied, the last time he took his feet down he made me a million dollars.
You can tell if people are looking busy.
You can tell who you like.
You can tell who is the most sexually attractive.
Measuring the productivity of highly intellectual people is just not possible Mark.
How are you going to measure the performance of an operation that's never been done before. It is going to be based on this quarter's results or this years results?

[Mark Herschberg]

Originally posted by Rufus BugleWeed:

If a guy delivers a solution faster but has more rework and maintenance associated with it was he really more productive?

I was expecting this. The answer to the above is "I don't know." That's a tough one, and extremely hard to answer in the abstract. but just because you don't know how to measure something doesn't mean it's not measurable. More specifically, just because I can't measure in this one case does not mean I can never measure.
Consider the following. There are a number of programming contest In high school, for example, I used to do the annual ACM contests. The finals went as follows. You lock 5 developers in a room with 3 computers for 5 hours. You give them 5 problems (I recall one was a variation on the traveling salesman problem, one year). The program is supposed to produce a certain output, e.g. optimal solution to the traveling salesman problem. At the end of 5 hours, all of your programs are run. You get one point for each problem you get right.
(Note: the ACM scoring was slightly differing, I think, and involved multiple answers printed out, allowing for partial credit, and there was also some bonus for finishing problems early. In any case, let's just take the above as the scoring system.)
Team A gets 4 problems done in 5 hours. Team B does 1 problem. Team A is more productive. (Repeat 100 times if you think it could be "luck" and pick teams with those averages.)
You can claim this isn't an accurate measurement, but you can't claim it's *totally* arbitrary and off-base. The expected output is like requirements and time is, well, time. Certainly we've ignored issues like maintanability, test coverage, and documentation. In theory, your program could just print out the final answer and "get lucky." We ignored code size, portability, and deployability. The requirements of the system are extremely simplified, i.e. print the solution and you're either right or wrong. Nevertheless, we can measure the productivity of these developers and come up with a ranking.
The real world is much much harder. Our models, be it SLOC, or bug count, or reuirements completed/hour are very rough gagues of productivity, but they are gagues. Bottom line: we can measure productivity.

Originally posted by Rufus BugleWeed:
What are the units?

Addressed above.
Recommended reading:
Peopleware (they've been running producitivyt test for years)
Measuring and Managing Performance in Organizations

--Mark
[ January 14, 2003: Message edited by: Mark Herschberg ]

[Rufus BugleWeed]

Even in this contrived case you are comparing the productivity of Team A vs Team B.
Development is not like hockey where we can record shots on goal, points scored per player, or number of penalty minutes.
Managemenent cannot measure productivity. Management knows they have a budget. They go out and get the cheapest people they can. Because they cannot get a good grip on quality either. They hope for the best.
With the cheapest people they feel they will most likely underspend their budget and get a bonus.
If they hire higher quality people they fininsh the requirements and add more. But the manager is less likely to get a bonus.

[Mark Herschberg]

Originally posted by Rufus BugleWeed:
Even in this contrived case you are comparing the productivity of Team A vs Team B.

So what? You claimed productivity was unmeasureable. I showed you that it was. Now you're complaining that you meant "objecttive productivity." How do you tell which airline pilot is the best? How do you tell which artist is better, or writer? What about CEOs and managers, or military officers. In all cases, we can define better and worse. Often there is not an objective measurement, mecause no two situations are the same.

Originally posted by Rufus BugleWeed:
Managemenent cannot measure productivity.

I just showed you how. Read the books I mentioned. read quality is free. In the worst case, take two sets of programmers, give them the same task and same resources, and same managers and see which makes more money. That will tell you who is more productive. Is this easy to do? Not at all. But you claimed measurement is impossible and that is not true. Difficulty != impossible.
Heck, given two candidates for a programming job, if you can't tell who will be more productive how do you hire one other then using a coin flip?

Originally posted by Rufus BugleWeed:

Management knows they have a budget. They go out and get the cheapest people they can. Because they cannot get a good grip on quality either. They hope for the best.
With the cheapest people they feel they will most likely underspend their budget and get a bonus.

Now you're just stereotyping. Its unfortunate that this is your view/experience with managers. I've worked for good ones who recognize quality and can distinguish better programmers from worse programmers.
IBM, Sony, Fujitsu, Apple, etc are all companies which recognize the differing abilities of people and reward appropriately. Read Quality is Free.

--Mark

[Rufus BugleWeed]

No, no, no
Your case above shows two potential employees and shows that one after x months has been some number and the other after the same period of time has been some other number. The implication is that the higher number is the better choice. The numbers are meaningless.
The numbers are just as meaningless as comparing two candidates grade point averages. But these numbers give a person or a judge a warm fuzzy feeling.
Flipping a coin may work as well as the way people make hiring choices for technical people.
IBM and Apple have both squandered their positions as market leaders. Those kind of companies gravitate to hiring the best looking candidates with the highest GPA from an elite selection of schools. When their new crop gets to the plate the Bad News Bears are inclined to wip them.
Small caps beat the large caps because the small caps can't afford to do what they believe is right and sometimes they happen to flip the coin heads 5 times in a row.
I'm not disputing some managers, technical workers, or Big Macs are better than others.
My point is you cannot measure the productivity of a high tech worker.
If quality is free why does a Toyota cost more than GM?