Der nachfolgende Text wurden mit KI erstellt und kann Fehler enthalten. Fehler gefunden? Bei GitHub editieren

IT im Jahr 2034 – Wo wollen wir hin? …collected on flipcharts and on notes.

We’ll start with the first question.

I always read out an answer and the two gentlemen can discuss something.

If I come up with something clever, I’ll just discuss it.

We’ll start with the question.

What challenges should we have solved in IT by 2034?

The first answer, which was also mentioned twice, is green IT.

I also count energy demand and energy consumption.

It was mentioned several times.

Do you have an opinion?

I would say that it is quite obvious.

With the use of cloud technologies and especially now with Gen AI, we have the big discussion that we have high electricity consumption and green IT is definitely an issue.

Exactly, so I would also write that down.

We actually had an episode on the subject not too long ago.

I took that as a bottom line.

Well, it’s good if you use cloud because of the higher packaging density.

That’s something that plays a role.

To be honest, from my point of view, this is unfortunately a socio-political issue.

What I read in the CT editorial a long time ago, half a year, three quarters of a year or so, is that if we go through in Germany to introduce a speed limit of 100 kmph on the motorway, then that saves as much as all the data centers that we have in Germany.

To be honest, I didn’t research that.

That’s one of the to-do tasks that I still have with me.

If that’s the case, then we have a different topic.

Then green IT is not the solution to this challenge.

It is definitely a challenge, but it is a socio-political problem.

To be honest, I find it difficult at this point because there is such an effect.

How should I put it?

We can fly less now.

We can do whatever we want.

But if such measures, for example, have these consequences, then we have a political issue.

I think that’s an important point.

It’s a bit like what we discussed in this episode.

I think it’s important and good that we take responsibility for our society and do something about it.

Just like you said, AI also ensures that we create new capacities over time.

But that’s a broader topic.

I think we have to act more broadly.

Do you already have an idea of what each of us could do to make it better?

In the direction of green IT?

Spontaneously, but you can try.

I think there are quite a few levers.

I just came up with this rebound effect.

If the processors are more powerful and consume less energy, then you may consume more energy because you say to yourself, well, it’s not that bad.

You should also pay attention to that.

But what everyone can do is, I think, if we do DevOps here and, I don’t know, five deployments a day and run all the tests every time, drive up and down the environment very elaborately, you can think about whether one deployment a day or a week is enough.

That can save a lot, even if you take into account scale effects.

If an image is delivered in the right size, then it may be relatively little energy that is saved on one request.

But if this page now has millions of calls, then it sums up very quickly.

My recommendation would be the episode with Aydin on the topic of green IT, which we did not do for too long.

I think it was very helpful.

We had another episode, which I can’t remember, with Martin Lippert and Stefan Roku.

We also talked about the topic of sustainability.

There were a few other aspects.

Stefan’s employer, after IT Agile, sat down and said, let’s see what that means for us as a company.

Then things like the way we do IT training, for example, which is obviously an important topic for them, leads to higher or lower sustainability.

We now have two episodes in the advent calendar.

The very first, where Jutta presents a book about sustainability, green IT, and Aydin, who was with me a few days ago, also has a book about practical green IT stories.

These are the approaches I would see.

For me, there is not much to say about the content itself.

It’s just not my deep expertise.

I found it exciting.

I also think it’s cool that you made an assessment in that direction.

What I find interesting again is that the green IT of electricity consumption is actually linked to the costs via the cloud resources.

If you reduce electricity consumption, the cloud costs will drop.

It should definitely be a goal of every company to reduce costs.

It goes hand in hand and just makes sense.

Very nice.

Let’s move on to the next thing that was mentioned.

What challenges should we have solved in IT by 2034?

Here is a bubble.

I’ll read out everything that comes to mind.

Encryption, quantum, dot, dot, dot, cryptography and quantum computing, mentioned by Ralf D.

Müller, and making quantum computing possible for the end consumer, mentioned by Phishen3.

If you have already read my name, I find it exciting when quantum computing comes up.

Then it is always said that we get problems with cryptography, that some algorithms no longer withstand the attack on quantum computing.

To be honest, I have no idea about quantum computing.

But I see it as a challenge that it will come.

My perception is that there are mechanisms, quantum, computing-safe cryptography to operate.

But I don’t remember the specific technologies.

At this point, an anecdote, because we are currently facing challenges.

The United States of America, the people there in the field of law enforcement, are now of the opinion that cryptography and private communication is actually something important.

And that it is also something that is protective and helps American national security in this case.

And that’s because they obviously have Chinese hackers who have hacked the system for monitoring telecommunications.

And that’s why the telecommunications in the USA can stop.

And I think that’s a positive development in that sense.

Well, it’s obviously not a negative development, but it has a positive aspect.

It has the positive aspect that we can now say live that encryption of information is a fundamentally positive thing because it helps exactly against such things.

And data storage, which says we can stop everything, for example, through law enforcement, leads to a very valuable goal, which other nations very much like to attack.

And in this case, they have successfully attacked.

So maybe that’s a good idea to do something like that.

So I only read a headline in which it was said according to the motto, yes, American law enforcement thinks cryptography is good as long as there is a backdoor in it that they can use.

But that’s exactly the point.

So if you have this backdoor, and they have now proven this experimentally, you create a goal with it that other people can take advantage of, for example, the Chinese in this case.

By the way, I find that very interesting again, because we now have a technology, Gen AI, sorry that I’m already getting to it.

It happens quite often, by the way.

Overcomes language barriers.

And that’s why I find it interesting that these two groups, the Chinese, as you say, it doesn’t light up for me.

State-level actors who are not friendly to you.

Okay, let’s go to the next point.

I know that one statement from Eberhard is already happy.

One is, so what challenges do we want to solve?

I remember again, someone wrote constantly increasing complexity and someone else wrote decreasing increasing complexity.

The funny thing about it is that decreasing increasing complexity is a contradiction in itself.

So either it increases or it decreases, but you can’t have both at the same time.

I don’t know how you see it.

I would argue that this is a bit of a goal that we may always have and write on the flags.

I had to think of a talk that I gave a while ago on the topic, don’t we actually ask for complexity?

And there is also this beautiful saying that technicians simply raise problems to a level of complexity, where it actually challenges them seriously.

And so I would say, it is obviously a meaningful demand.

This is perhaps a traditional demand, a traditional challenge.

But on the other hand, it may also be the case that we have to look each other in the eye and say, we might like to deal with non-trivial problems and that’s why complexity may be a good thing.

I would like to ask a quick question in the room.

Has any of you ever seen in the code that a really simple thing was solved very, very, very cumbersome and complex, just to see if it works?

If no hand goes up now, I know you don’t listen to me.

Yes, right?

Hands.

Everyone doesn’t dare.

You’re talking about the Java certification?

Possibly.

And from other things.

I think the quote comes from Neil Ford.

Developers are attracted by complexity, like moths by light, usually with the same result.

They burn.

What’s going on there?

I think that’s actually quite interesting, because it resonates with developers.

They agree with it.

And that just leads to very interesting things.

Yes, you don’t really want the complexity, because you know that at some point you no longer master your own complexity.

Because debugging is a bit more complex than writing the code.

And when I get to my limits with writing the code, then I can no longer debug.

And also the work in the team suffers extremely from it.

Because if you understand the code, it’s difficult for everyone else.

And I’m just wondering, we already had green IT at the beginning and reducing costs caused by energy.

Does that sometimes play a role for you?

That’s going to be really exciting now.

Because I was looking for a way to best read a QR code.

And I found a lot of blog posts.

Yes, I trained a model there.

And then I thought, okay, that’s kind of the wrong approach.

And right now with Gen AI, yes, I can just throw in a QR code, get the result out.

Has the complexity been lowered for me as a developer?

But it’s not about green IT.

Yes.

Oh, okay.

Eber shields already.

He wants to continue.

Okay.

That sounds like …

Sorry.

Exactly.

I had to pick up the idea again, so to speak.

I think it’s clear that we’re actually talking about a misdevelopment.

So actually, I would expect that developers at a certain seniority level are on this level of, I’m doing something boring, which solves the problem, and I don’t make it extra complicated.

But unfortunately there is this other trend.

And I think you have to say that somehow, if you say, well, we want to reduce complexity.

Because if you stay there, you put it on hold that there is the opposite thing, that people are trying to increase complexity.

And if we don’t realize that, we don’t know how we can reduce it.

Because then we don’t tackle the social problem, but somehow find purely technical solutions.

And that’s not the solution.

Obviously, when people actually say, yes, complexity is totally great, and maybe they don’t say that explicitly, but behave that way, then the solution is not to say, here is a technical solution for your complexity problem, but then it is the solution to ensure that it is implemented.

And that’s another solution strategy.

But first you have to understand what the problem is at all.

And that’s why I think this reference to this social problem is important.

I have the feeling that every episode of Software Architecture in the stream summarizes what it used to be.

First you have to understand what the problem is, and then you can develop the right solution based on it.

But I don’t want to reduce Software Architecture in the stream.

That’s also a life hack, isn’t it?

Yes, that’s right.

I’m looking at Ralf inconspicuously now, because I’m sure he wants to say something about the next one.

What challenges do we want to solve by 2034?

Solve data protection problems with AI.

You go first.

The topic has your name on it, so to speak.

Just because AI is on it, I don’t have to be in it.

Data protection with, let’s say, Gen AI is a very diverse problem.

If we now look at person-related data, then hopefully we only have data from public data.

But as the problem is currently being played, so with ChatGPT, a few names are now simply blocked.

If you ask the model to give out the name, then the discussion simply breaks down.

It is assumed that this is due to our data protection regulations, that the model would otherwise give out wrong data, wrong information about the user.

And that’s why this block is there.

I think it’s a very, very difficult topic to answer briefly now.

So we should write a note.

That could lead to a whole episode, because if so, I’ll do it now.

I’m not sure.

I don’t quite understand the relationship between the two problems.

Data protection, as well as copyright laws, are universal.

And what we see in copyright law is that there is this statement that copyright must be dissolved or abolished by AI in the future.

And that is absurd to a large extent.

On the other hand, you have to deal with massive penalties if you just copy a movie.

And maybe this is a similar topic here.

These are fundamental things.

I don’t quite understand what the relationship with AI should be at this point.

You have to look at it in principle.

And it is actually so, to put it positively, I found these stories about GPID and data protection actually very positive, because I noticed in my consulting work that the topic is really taken seriously and is actually implemented.

And I think that’s actually positive.

So that’s why it’s something where we have things at certain points that I think work well.

It’s positive, yes.

But it’s very, very difficult to implement in terms of Gen AI and training data sets.

Because if I ask for a person’s birth date, yes, Ralf Müller, that’s a collective term.

Sure, I’ll get something wrong if the machine gives me something wrong.

Not my birth date.

And that’s why it’s then again, if the name is now blocked.

So this topic just goes very deep.

And then you have to think about it.

I don’t know if the question was actually just data protection.

So protection of personal data, as it is interpreted here in Germany.

Or whether it also goes in the direction of copyright.

And also there.

But you’re sketching a different problem.

Namely the problem that this system gives some answers that are not correct.

That’s something else.

And something that we discussed together.

Just now, when was it?

Last week or the week before last.

Where we found that just these results, that you have to look at them critically.

Data protection would be something else.

I don’t know if I should be honest.

We have approaches with which we can use medical data to conduct scientific research, for example.

Medical data is highly protective.

That works, yes.

Why should that be a different topic for AI now?

It’s the same topic.

Whether I write a paper and use medical data or whether I train an AI for it.

Same topic, right?

Data protection goes in several directions.

So once that my personal data has been leaked somewhere, especially medical data.

But also that they are correct.

I have a right to that.

And that’s actually quite good that it has been implemented.

That if false data can be found on the Internet about me, that I can delete them.

Can be deleted.

And I can’t do that with AI anymore.

Or if I delete something with such a general name.

And that doesn’t even occur to many people.

But with a general name like the one I have, data is very, very easily mixed up.

You somehow look for the name and just find the wrong person.

And that too belongs to data protection.

Okay.

I notice a complex topic.

Let’s move on.

I look at Ralf again inconspicuously.

Have fun watching, Ben.

AI as a matter of course.

Again from Fischen3.

By the way, I think the name is funny.

Here, too, I think we have these two differences again.

Simply AI, Machine Learning and Gen AI.

And I think a lot of people are now saying that the handling of Gen AI, with systems like Chatship-T, Cloud AI and so on, will become a matter of course.

And in many areas we won’t even notice where everything, including Machine Learning and other AI systems, is in use.

This also simply improves the human-machine interface, because you can speak to the machine in a natural language and don’t have to click through any settings deserts or things like that.

I don’t think I understand the challenge.

So what you’re saying is it’s going to be a matter of course, but it’s not a challenge.

It’s a prediction of a technological evolution.

Definitely.

I don’t see it as a challenge either, but it will turn out that way.

Then let’s continue inconspicuously with GPT garbage.

I liked it.

I recently learned a term for it.

A question to the audience.

Who has been looking for Baby V lately?

So the bird.

In the meantime, you can find a lot of pictures of…

It just doesn’t make sense.

Nobody would have come up with the idea of equipping light bulbs with Wi-Fi.

Exactly, and so on.

So that doesn’t mean that light bulbs now have software.

So, crazy, obviously crazy.

But that means, I would put in the room, but you can discuss it, that is, if we have such a boost in productivity, which, in quotation marks, only leads to the fact that we somehow have more software, and that is somehow this rebound effect.

You just mentioned it, right?

So, where I say, okay, something is becoming more efficient.

So, I have cars that consume less fuel.

The fleet consumption does not go down, but grows, because more people drive more cars, because it has become cheaper.

More people develop more software, because it has become cheaper.

And maybe that’s what’s in the house for us.

I have no idea.

But that’s something we’ll discuss next week, over next week.

On the 20th?

Next week, yes.

And then we might have a security problem again.

If we have ten times as much software, then we probably also have ten times as many security gaps, ten times as many systems that need to be patched.

We can’t do that with the ten people, but we have to automate it and it’s going to be exciting.

So, maybe briefly.

We actually planned a stream next Friday, where Stefan Schmidt, Andre Neubauer, Ralf and I, with Lisa as moderation, discuss exactly this topic, whether AI is somehow under or overhyped.

And that’s a bit of a preview, I think, whether that’s the case or not.

What I think is quite funny, which challenge should be solved.

You said you were able to screw together a single board thing.

Ewa said it’s starting to get tricky.

Basic knowledge of physics was mentioned, and I would throw it into the room so that it certainly implies a bit of electrical engineering.

And I just found it exciting that it was mentioned as a challenge that should be solved.

I actually studied computer science and chose physics as a secondary subject.

And that’s one of my regrets, because I should have taken philosophy, because on the one hand, it would have interested me more, as I found out afterwards, and on the other hand, because many of the things we do have to do with completely different effects.

So we have concept education, we create concepts, we write them in code, that has to be clear, that’s a philosophical thing.

And we also have this problem with these people, who develop software, who set requirements, and so on.

That’s why I would say, I would like to have basic knowledge in social or intellectual sciences.

And I think that’s actually a luxury.

I really miss it.

I don’t understand why I need physics.

I hardly ever needed it in my professional life, just like I needed theoretical computer science, for example.

So it’s a really exciting thing.

I found it great at the university, this whole complexity thing.

So it doesn’t play a role for me, and it doesn’t play a serious role for the problems I’m trying to solve.

I’m thinking about it a bit.

There is this paper, how AI will scale in the future, how you calculate from the past, how AI has scaled, how it will continue to scale.

And if you look at it, the paper talks about how we will reach general AI very quickly.

But what is not taken into account are the physical limits, which then also, especially in terms of energy consumption, if it continues to scale, reach the planetary limits.

And of course that’s also physics.

So you should think about it somewhere, that limits are set to the whole.

Yes, but it’s also a topic, I don’t even know how that became clear to me.

If you assume unlimited exponential growth, many things are fantastically easy to solve.

But that’s completely unrealistic.

And that’s a bit what you’re saying, okay, there are limits.

But I think that’s a fundamental problem on many levels, and I think it’s a growth problem.

And it’s hard to convey, or in some places it’s hard to convey, that it just doesn’t work.

Because if I have unlimited exponential growth, at some point it’s so incredibly much that it goes wrong.

Let’s go to the next one.

I would like you to not fall asleep here.

Did any of you see, what is the most important skill in IT in 2024?

No, no, anyone?

Okay, no, not yet, but maybe next year again.

Which challenge should we have solved by 2034 was mentioned several times as what was the most important skill that came out?

Does anyone know?

Prompt engineering.

Unfortunately, no.

Communication, communication.

But we definitely have prompt engineering at some point, but in a different context.

It’s also part of communication.

Yes, exactly, it’s more important that you can speak with chat GPT than with us.

I didn’t say more important, it’s a subcategory of communication.

We have human communication and machine communication.

Is it possible to solve the challenge in 10 years with this communication?

If we understood it as a problem at first, I think that’s worth a lot.

I find it difficult in some places, because I have the feeling that the level of communication and the problem is unclear to me.

I could refer to this article from the Heise Forum, where it actually said about one of our streams, I forgot which one it was, where it said that the communication problem is a reason why we should introduce defense service again.

And that’s because a standardized communication is transmitted in a standardized way.

At least two in the audience are laughing.

How should I say it?

When I communicate and gesticulate and say something, and how I emphasize it and so on, there is a depth of communication and a diversity that contradicts the concept that is shimmering through so blatantly.

The perception of what communication is and we can communicate without talking.

I’ve heard that’s also possible.

And if you say that’s a solution, then it means that the whole problem statement is completely unclear.

And then we have another problem.

Then it’s unclear what communication is.

And there we are again with this spiritual thing.

If I understand what language is, then maybe I’ll get clearer with the communication problem.

That’s why I say spiritual-scientific foundations.

But let’s not try to solve a large part of the communication problem via feedback loops.

Especially with Agile, for example, that we develop little by little, show the result, ask, is it the right thing, is that what you tried to say, no, I wanted a little further right, a little further left?

The problem is, I can make a statement and mean the opposite.

And that depends on an intonation, it depends on whether it is sarcasm, sarcasm may not be recognizable, and so on and so on.

And that’s what I’m trying to say.

Communication is somehow more diverse and is a different topic.

There are also other things involved.

Another example is this story with 40 people or so.

Nobody said anything.

I suspect that many people have a lot of feedback, but it’s this problem with hello, I would like to say something too.

And that means we have a communication problem.

We have an inherent communication problem because the audience is the audience and to a certain extent is condemned to passivity.

And you have to realize that.

That is, if I would stand here and say, I have told you various things.

I have at least or we have told that we have at least not provoked a scream.

But it can be very good that all 40 people who are sitting here think that we are telling complete nonsense.

I don’t know.

By the way, someone has signed up.

Do you want to say something about it?

Wait, I’ll bring you a microphone.

Do you want to …

So there you are right in the camera.

You can also be the voice from the off.

As you want.

Hopefully, otherwise I’ll give you mine.

Try again.

No, it’s not on.

Get mine, then I’ll scream at them.

Yes, maybe with communication an interesting point that this is now through society but very strongly polarized.

That is, the challenge of communication in recent years was perhaps the complexity or the difficulty to communicate at all or to communicate better.

But through the media or through society, I think, communication will be so difficult in the next few years or partly so difficult that there is a risk that the challenge will increase.

Or didn’t you say?

So I can so to speak from a personal perspective without having a high superstructure answer for myself.

And there are different things for me.

So if someone tells me here is a solution for a technical problem not a typical consulting thing then I start and talk to this person and if I have the feeling that this technical solution is absolute nonsense then I have no problem with it but it’s something that happens more often and we can communicate rationally about it because in the end it’s about solving a technical problem.

That’s something completely different when it comes to certain political statements that contradict my conviction.

And that’s what we actually have with social polarization.

We have a social polarization in political terms and I think I personally believe that in most cases you can separate it from this technically rational discussion.

Because if you tell me here is a technical solution and that’s absurd then I can deal with it and we can have a discussion about what the better technical solution is.

Certain political statements that we have in society are completely different and at least trigger me on a different level.

Yes, maybe still if I talk to people nowadays and ask for a solution I often get one solution or one alternative only.

So that means the multiplicity or the variability in the answer I do believe that it has subsided or I have the feeling when I talk to the younger people that they don’t see these 2-3 solutions or these different alternative ways but only this one problem or solution.

On a technical level I would suggest that it depends on the seniority.

There is the seniority level where you say there is a technology X that I find that solves all the problems and I don’t understand why people still have mainframes.

And at some point you realize that there are many solutions and that there are different solutions for technical problems and that it’s a permanent game of trade-offs.

In my eyes I believe it’s a seniority problem and for me it’s fine.

That’s why we make experiences.

I believe it’s something different than in this non-private political discussion.

That’s my claim.

Which challenges should we have solved?

Server-side rendering should be implemented again.

Thank you for the courage.

A round of applause.

Server-side rendering should be implemented again.

Single-page apps should be gone.

I just wanted to add something on a technical level.

I know there is a lot in the audience that wanted to be discussed but it just fit so well that I wanted to add it.

I don’t think you want to say anything because you are not the front-end.

I read an article this week that said that in view of something like Apple Vision Pro and VR the browser as a universal front-end is a problem.

We talked about this in our last stream.

The question is whether the browser as a universal front-end in view of this AI topic is still relevant because certain things become machine-to-machine communication.

I don’t know.

Front-ends are gone.

I have a Mac front-end.

I have a Windows front-end.

I have the browser front-end as a universal front-end.

Maybe we see a change through such a mega-transfer like AI.

Then this question about server-side rendering and SPA is no longer a serious question because we have a different topic.

I would like to say the next thing because it affected me recently.

What challenges should we solve?

Full-scale compatible calendar systems.

You don’t hear this in the stream, but the audience laughs.

I was very happy.

Here was the example that single appointments can be handled with term series deleted.

In practice, daily appointments from Google Calendar are not a problem because everything is different.

It’s great.

Do you think we solved this in 10 years?

Even Eberhard had to smile.

Since we even have two time zones in Germany, this is really a big problem.

These are two time zones that are synchronized.

You can read it.

This whole calendar thing is a real advantage if we don’t communicate.

We don’t need to synchronize appointments anymore.

Then you don’t need to solve it anymore.

This is in a way a reference to the general phenomenon.

We’re talking about AI developers becoming superfluous and we have established job interviews that say, solve this complicated problem.

Then it’s just that such stories in two time zones in Germany lead to me needing a week to find out why this stupid mistake is there.

I can’t explain to anyone why it took a week which is very low-level work.

This is almost schizophrenic.

I think we all know this.

You have this little thing which is insanely complicated and if you stand like this and discuss the big pictures then it gets a bit in the background.

I think this is an example.

It’s a seemingly trivial topic.

We could completely switch to asynchronous communication and then we wouldn’t need the synchronization via calendar anymore.

It’s a good example that software architecture means to think about different solution alternatives and to realize which different trade-offs there are.

Asynchronous communication would be a solution.

I think it’s a good example that in technical discussions certain statements don’t trigger me as much as in other discussions.

I just said through the flowers that this idea is shit.

It would have been really nice.

I said it again and I meant it.

I find it creative to some extent.

Seriously.

This meeting is now an email because we have a calendar system with which we can’t organize meetings.

Please jump to the next one.

The next one is about software quality and you have to leave.

How long do we want to do this?

An hour?

We’ll talk about software quality later.

Ah, okay.

Software quality is an important architecture driver, I would say.

And it remains an important topic.

Obviously.

Ralf, would you like to say something about software quality?

I’m still trying to catch myself.

Software quality is definitely important.

It’s a serious topic, isn’t it?

It’s a driver for architecture.

By the way, we’ve done an hour now and we still have 20 solutions.

I can’t continue to motivate you.

We’ll do another follow-up stream, okay?

I’d love to.

I’m really sorry.

First of all, thank you to everyone who watched online.

Thank you to everyone who still wants to listen and watch the recording.

Thank you to you in the audience who were there, who talked with us.

Thank you very much.

I also thought it was cool that you called in.

Eberhard, is there a stream on Friday?

Not at all.

But on the 20th.

Eberhard has already advertised for it.

If you can’t make it on the 20th, decorate the Christmas tree.

Look at the software architecture in the recording.

I think that’s pretty cool, too.

That’s actually the last one for this year.

I guess we’ll continue next year.

I don’t know if we’ll make it on the first Friday, but we’ll see.

Many, many thanks to you all.

It was great.

Thank you to everyone who was there.

Have a wonderful time here on the IT days when you are live on site.

Enjoy the time.

See you soon.

Thank you.