5 VC investment criteria

I reviewed 200 startup business concepts last week. Brainache.

The enterprises sold various goods and services. The concepts were achingly similar: give us money, we'll produce a product, then get more to expand. No different from daily plans and pitches.

Most of those 200 plans sounded plausible. But 10% looked venture-worthy. 90% of startups need alternatives to venture finance.

With the success of VC-backed businesses and the growth of venture funds, a common misperception is that investors would fund any decent company idea. Finding investors that believe in the firm and founders is the key to funding.

Incorrect. Venture capital needs investing in certain enterprises. If your startup doesn't match the model, as most early-stage startups don't, you can revise your business plan or locate another source of capital.

Before spending six months pitching angels and VCs, make sure your startup fits these criteria.

Likely to generate $100 million in sales

First, I check the income predictions in a pitch deck. If it doesn't display $100M, don't bother.

The math doesn't work for venture financing in smaller businesses.

Say a fund invests $1 million in a startup valued at $5 million that is later acquired for $20 million. That's a win everyone should celebrate. Most VCs don't care.

Consider a $100M fund. The fund must reach $360M in 7 years with a 20% return. Only 20-30 investments are possible. 90% of the investments will fail, hence the 23 winners must return $100M-$200M apiece. $15M isn't worth the work.

Angel investors and tiny funds use the same ideas as venture funds, but their smaller scale affects the calculations. If a company can support its growth through exit on less than $2M in angel financing, it must have $25M in revenues before large companies will consider acquiring it.

Aiming for Hypergrowth

A startup's size isn't enough. It must expand fast.

Developing a great business takes time. Complex technology must be constructed and tested, a nationwide expansion must be built, or production procedures must go from lab to pilot to factories. These can be enormous, world-changing corporations, but venture investment is difficult.

The normal 10-year venture fund life. Investments are made during first 3–4 years.. 610 years pass between investment and fund dissolution. Funds need their investments to exit within 5 years, 7 at the most, therefore add a safety margin.

Longer exit times reduce ROI. A 2-fold return in a year is excellent. Loss at 2x in 7 years.

Lastly, VCs must prove success to raise their next capital. The 2nd fund is raised from 1st fund portfolio increases. Third fund is raised using 1st fund's cash return. Fund managers must raise new money quickly to keep their jobs.

Branding or technology that is protected

No big firm will buy a startup at a high price if they can produce a competing product for less. Their development teams, consumer base, and sales and marketing channels are large. Who needs you?

Patents, specialist knowledge, or brand name are the only answers. The acquirer buys this, not the thing.

I've heard of several promising startups. It's not a decent investment if there's no exit strategy.

A company that installs EV charging stations in apartments and shopping areas is an example. It's profitable, repeatable, and big. A terrific company. Not a startup.

This building company's operations aren't secret. No technology to protect, no special information competitors can't figure out, no go-to brand name. Despite the immense possibilities, a large construction company would be better off starting their own.

Most venture businesses build products, not services. Services can be profitable but hard to safeguard.

Probable purchase at high multiple

Once a software business proves its value, acquiring it is easy. Pharma and medtech firms have given up on their own research and instead acquire startups after regulatory permission. Many startups, especially in specialized areas, have this weakness.

That doesn't mean any lucrative $25M-plus business won't be acquired. In many businesses, the venture model requires a high exit premium.

A startup invents a new glue. 3M, BASF, Henkel, and others may buy them. Adding more adhesive to their catalogs won't boost commerce. They won't compete to buy the business. They'll only buy a startup at a profitable price. The acquisition price represents a moderate EBITDA multiple.

The company's $100M revenue presumably yields $10m in profits (assuming they’ve reached profitability at all). A $30M-$50M transaction is likely. Not terrible, but not what venture investors want after investing $25M to create a plant and develop the business.

Private equity buys profitable companies for a moderate profit multiple. It's a good exit for entrepreneurs, but not for investors seeking 10x or more what PE firms pay. If a startup offers private equity as an exit, the conversation is over.

Constructed for purchase

The startup wants a high-multiple exit. Unless the company targets $1B in revenue and does an IPO, exit means acquisition.

If they're constructing the business for acquisition or themselves, founders must decide.

If you want an indefinitely-running business, I applaud you. We need more long-term founders. Most successful organizations are founded around consumer demands, not venture capital's urge to grow fast and exit. Not venture funding.

if you don't match the venture model, what to do

VC funds moonshots. The 10% that succeed are extraordinary. Not every firm is a rocketship, and launching the wrong startup into space, even with money, will explode.

But just because your startup won't make $100M in 5 years doesn't mean it's a bad business. Most successful companies don't follow this model. It's not venture capital-friendly.

Although venture capital gets the most attention due to a few spectacular triumphs (and disasters), it's not the only or even most typical option to fund a firm.

Other ways to support your startup:

• Personal and family resources, such as credit cards, second mortgages, and lines of credit

• bootstrapping off of sales

• government funding and honors

• Private equity & project financing

• collaborating with a big business

• Including a business partner

Before pitching angels and VCs, be sure your startup qualifies. If so, include them in your pitch.

My ego is so huge it won't fit through the door.

I don't know how I feel about it. I should be excited. Many of you have this exact dream to publish a book with a well-known book publisher and get a juicy advance.

Let me dissect how I'm thinking about it to help you.

How it happened

An email comes in. A generic "can we put a backlink on your website and get a freebie" email.

Almost deleted it.

Then I noticed the logo. It seemed shady. I found the URL. Check. I searched the employee's LinkedIn. Legit. I avoided middlemen. Check.

Mixed feelings. LinkedIn hasn't valued my writing for years. I'm just a guy in an unironed t-shirt whose content they sell advertising against.

They get big dollars. I get $0 and a few likes, plus some email subscribers.

Still, I felt adrenaline for hours.

I texted a few friends to see how they felt. I wrapped them.

Messages like "No shocker. You're entertaining online." I didn't like praises, so I blushed.

The thrill faded after hours. Who knows?

Most authors desire this chance.

"You entitled piece of crap, Denning!"

You may think so. Okay. My job is to stand on the internet and get bananas thrown at me.

I approached writing backwards. More important than a book deal was a social media audience converted to an email list.

Romantic authors think backward. They hope a fantastic book will land them a deal and an audience.

Rarely occurs. So I never pursued it. It's like permission-seeking or the lottery.

Not being a professional writer, I've never written a good book. I post online for fun and to express my opinions.

Writing is therapeutic. I overcome mental illness and rebuilt my life this way. Without blogging, I'd be dead.

I've always dreamed of staying alive and doing something I love, not getting a book contract. Writing is my passion. I'm a winner without a book deal.

Why I was given a book deal

You may assume I received a book contract because of my views or follows. Nope.

They gave me a deal because they like my writing style. I've heard this for eight years.

Several authors agree. One asked me to improve their writer's voice.

Takeaway: highlight your writer's voice.

What if they discover I'm writing incompetently?

An edited book is published. It's edited.

I need to master writing mechanics, thus this concerns me. I need help with commas and sentence construction.

I must learn verb, noun, and adjective. Seriously.

Writing a book may reveal my imposter status to a famous publisher. Imagine the email

"It happened again. He doesn't even know how to spell. He thinks 'less' is the correct word, not 'fewer.' Are you sure we should publish his book?"

Fears stink.

I'm capable of blogging. Even listicles. So what?

Writing for a major publisher feels advanced.

I only blog. I'm good at listicles. Digital media executives have criticized me for this.

• It is allegedly clickbait.

• Or it is following trends.

• Alternately, growth hacking.

Never. I learned copywriting to improve my writing.

Apple, Amazon, and Tesla utilize copywriting to woo customers. Whoever thinks otherwise is the wisest person in the room.

Old-schoolers loathe copywriters.

Their novels sell nothing.

They assume their elitist version of writing is better and that the TikTok generation will invest time in random writing with no subheadings and massive walls of text they can't read on their phones.

I'm terrified of book proposals.

My friend's book proposal suggestion was contradictory and made no sense.

They told him to compose another genre. This book got three Amazon reviews. Is that a good model?

The process disappointed him. I've heard other book proposal horror stories. Tim Ferriss' book "The 4-Hour Workweek" was criticized.

Because he has thick skin, his book came out. He wouldn't be known without that.

I hate book proposals.

An ongoing commitment

Writing a book is time-consuming.

I appreciate time most. I want to focus on my daughter for the next few years. I can't recreate her childhood because of a book.

No idea how parents balance kids' goals.

My silly face in a bookstore. Really?

Genuine thought.

I don't want my face in bookstores. I fear fame. I prefer anonymity.

I want to purchase a property in a bad Australian area, then piss off and play drums. Is bookselling worth it?

Are there even bookstores anymore?

(Except for Ryan Holiday's legendary Painted Porch Bookshop in Texas.)

What's most important about books

Many were duped.

Tweets and TikTok hopscotch vids are their future. Short-form content creates devoted audiences that buy newsletter subscriptions.

Books=depth.

Depth wins (if you can get people to buy your book). Creating a book will strengthen my reader relationships.

It's cheaper than my classes, so more people can benefit from my life lessons.

A deeper justification for writing a book

Mind wandered.

If I write this book, my daughter will follow it. "Look what you can do, love, when you ignore critics."

That's my favorite.

I'll be her best leader and teacher. If her dad can accomplish this, she can too.

My kid can read my book when I'm gone to remember her loving father.

Last paragraph made me cry.

The positive

This book thing might make me sound like Karen.

The upside is... Building in public, like I have with online writing, attracts the right people.

Proof-of-work over proposals, beautiful words, or huge aspirations. If you want a book deal, try writing online instead of the old manner.

Next steps

No idea.

I'm a rural Aussie. Writing a book in the big city is intimidating. Will I do it? Lots to think about. Right now, some level of reflection and gratitude feels most appropriate.

Sometimes when you don't feel worthy, it gives you the greatest lessons. That's how I feel about getting offered this book deal.

Perhaps you can relate.

How did his money blind me?

Like most fledgling entrepreneurs, I wanted a mentor. I met as many nearby folks with "entrepreneur" in their LinkedIn biographies for coffee.

These meetings taught me a lot, and I'd suggest them to any new creator. Attention! Meeting with many experienced entrepreneurs means getting contradictory advice. One entrepreneur will tell you to do X, then the next one you talk to may tell you to do Y, which are sometimes opposites. You'll have to chose which suggestion to take after the chats.

I experienced this. Same afternoon, I had two coffee meetings with experienced entrepreneurs. The first meeting was with a billionaire entrepreneur who took his company public.

I met him in a swanky hotel lobby and ordered a drink I didn't pay for. As a fledgling entrepreneur, money was scarce.

During the meeting, I demoed the software I'd built, he liked it, and we spent the hour discussing what features would make it a success. By the end of the meeting, he requested I include a killer feature we both agreed would attract buyers. The feature was complex and would require some time. The billionaire I was sipping coffee with in a beautiful hotel lobby insisted people would love it, and that got me enthusiastic.

The second meeting was with a young entrepreneur who had recently raised a small amount of investment and looked as eager to pitch me as I was to pitch him. I forgot his name. I mostly recall meeting him in a filthy coffee shop in a bad section of town and buying his pricey cappuccino. Water for me.

After his pitch, I demoed my app. When I was done, he barely noticed. He questioned my customer acquisition plan. Who was my client? What did they offer? What was my plan? Etc. No decent answers.

After our meeting, he insisted I spend more time learning my market and selling. He ignored my questions about features. Don't worry about features, he said. Customers will request features. First, find them.

Putting your faith in results over relevance

Problems plagued my afternoon. I met with two entrepreneurs who gave me differing advice about how to proceed, and I had to decide which to pursue. I couldn't decide.

Ultimately, I followed the advice of the billionaire.

Obviously.

Who wouldn’t? That was the guy who clearly knew more.

A few months later, I constructed the feature the billionaire said people would line up for.

The new feature was unpopular. I couldn't even get the billionaire to answer an email showing him what I'd done. He disappeared.

Within a few months, I shut down the company, wasting all the time and effort I'd invested into constructing the killer feature the billionaire said I required.

Would follow the struggling entrepreneur's advice have saved my company? It would have saved me time in retrospect. Potential consumers would have told me they didn't want what I was producing, and I could have shut down the company sooner or built something they did want. Both outcomes would have been better.

Now I know, but not then. I favored achievement above relevance.

Success vs. relevance

The millionaire gave me advice on building a large, successful public firm. A successful public firm is different from a startup. Priorities change in the last phase of business building, which few entrepreneurs reach. He gave wonderful advice to founders trying to double their stock values in two years, but it wasn't beneficial for me.

The other failing entrepreneur had relevant, recent experience. He'd recently been in my shoes. We still had lots of problems. He may not have achieved huge success, but he had valuable advice on how to pass the closest hurdle.

The money blinded me at the moment. Not alone So much of company success is defined by money valuations, fundraising, exits, etc., so entrepreneurs easily fall into this trap. Money chatter obscures the value of knowledge.

Don't base startup advice on a person's income. Focus on what and when the person has learned. Relevance to you and your goals is more important than a person's accomplishments when considering advice.

Deep learning-based generative art is being researched. As usual, self-learning is better. Some models, like OpenAI's DALL-E 2, require registration and can only be used online, but others can be used locally, which is usually more enjoyable for curious users. I'll demonstrate the Stable Diffusion model's operation on a standard PC.

Let’s get started.

What It Does

Stable Diffusion uses numerous components:

• A generative model trained to produce images is called a diffusion model. The model is incrementally improving the starting data, which is only random noise. The model has an image, and while it is being trained, the reversed process is being used to add noise to the image. Being able to reverse this procedure and create images from noise is where the true magic is (more details and samples can be found in the paper).

• An internal compressed representation of a latent diffusion model, which may be altered to produce the desired images, is used (more details can be found in the paper). The capacity to fine-tune the generation process is essential because producing pictures at random is not very attractive (as we can see, for instance, in Generative Adversarial Networks).

• A neural network model called CLIP (Contrastive Language-Image Pre-training) is used to translate natural language prompts into vector representations. This model, which was trained on 400,000,000 image-text pairs, enables the transformation of a text prompt into a latent space for the diffusion model in the scenario of stable diffusion (more details in that paper).

This figure shows all data flow:

The weights file size for Stable Diffusion model v1 is 4 GB and v2 is 5 GB, making the model quite huge. The v1 model was trained on 256x256 and 512x512 LAION-5B pictures on a 4,000 GPU cluster using over 150.000 NVIDIA A100 GPU hours. The open-source pre-trained model is helpful for us. And we will.

Install

Before utilizing the Python sources for Stable Diffusion v1 on GitHub, we must install Miniconda (assuming Git and Python are already installed):

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-Linux-x86_64.sh
chmod +x Miniconda3-py39_4.12.0-Linux-x86_64.sh
./Miniconda3-py39_4.12.0-Linux-x86_64.sh
conda update -n base -c defaults conda

Install the source and prepare the environment:

git clone https://github.com/CompVis/stable-diffusion
cd stable-diffusion
conda env create -f environment.yaml
conda activate ldm
pip3 install transformers --upgrade

Download the pre-trained model weights next. HiggingFace has the newest checkpoint sd-v14.ckpt (a download is free but registration is required). Put the file in the project folder and have fun:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Almost. The installation is complete for happy users of current GPUs with 12 GB or more VRAM. RuntimeError: CUDA out of memory will occur otherwise. Two solutions exist.

Running the optimized version

Try optimizing first. After cloning the repository and enabling the environment (as previously), we can run the command:

python3 optimizedSD/optimized_txt2img.py --prompt "hello world" --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Stable Diffusion worked on my visual card with 8 GB RAM (alas, I did not behave well enough to get NVIDIA A100 for Christmas, so 8 GB GPU is the maximum I have;).

Running Stable Diffusion without GPU

If the GPU does not have enough RAM or is not CUDA-compatible, running the code on a CPU will be 20x slower but better than nothing. This unauthorized CPU-only branch from GitHub is easiest to obtain. We may easily edit the source code to use the latest version. It's strange that a pull request for that was made six months ago and still hasn't been approved, as the changes are simple. Readers can finish in 5 minutes:

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available at line 20 of ldm/models/diffusion/ddim.py ().

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available in line 20 of ldm/models/diffusion/plms.py ().

• Replace device=cuda in lines 38, 55, 83, and 142 of ldm/modules/encoders/modules.py with device=cuda if torch.cuda.is available(), otherwise cpu.

• Replace model.cuda() in scripts/txt2img.py line 28 and scripts/img2img.py line 43 with if torch.cuda.is available(): model.cuda ().

Run the script again.

Testing

Test the model. Text-to-image is the first choice. Test the command line example again:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

The slow generation takes 10 seconds on a GPU and 10 minutes on a CPU. Final image:

Hello world is dull and abstract. Try a brush-wielding hamster. Why? Because we can, and it's not as insane as Napoleon's cat. Another image:

Generating an image from a text prompt and another image is interesting. I made this picture in two minutes using the image editor (sorry, drawing wasn't my strong suit):

I can create an image from this drawing:

python3 scripts/img2img.py --prompt "A bird is sitting on a tree branch" --ckpt sd-v1-4.ckpt --init-img bird.png --strength 0.8

It was far better than my initial drawing:

I hope readers understand and experiment.

Stable Diffusion UI

Developers love the command line, but regular users may struggle. Stable Diffusion UI projects simplify image generation and installation. Simple usage:

• Unpack the ZIP after downloading it from https://github.com/cmdr2/stable-diffusion-ui/releases. Linux and Windows are compatible with Stable Diffusion UI (sorry for Mac users, but those machines are not well-suitable for heavy machine learning tasks anyway;).

• Start the script.

Done. The web browser UI makes configuring various Stable Diffusion features (upscaling, filtering, etc.) easy:

V2.1 of Stable Diffusion

I noticed the notification about releasing version 2.1 while writing this essay, and it was intriguing to test it. First, compare version 2 to version 1:

• alternative text encoding. The Contrastive LanguageImage Pre-training (CLIP) deep learning model, which was trained on a significant number of text-image pairs, is used in Stable Diffusion 1. The open-source CLIP implementation used in Stable Diffusion 2 is called OpenCLIP. It is difficult to determine whether there have been any technical advancements or if legal concerns were the main focus. However, because the training datasets for the two text encoders were different, the output results from V1 and V2 will differ for the identical text prompts.

• a new depth model that may be used to the output of image-to-image generation.

• a revolutionary upscaling technique that can quadruple the resolution of an image.

• Generally higher resolution Stable Diffusion 2 has the ability to produce both 512x512 and 768x768 pictures.

The Hugging Face website offers a free online demo of Stable Diffusion 2.1 for code testing. The process is the same as for version 1.4. Download a fresh version and activate the environment:

conda deactivate  
conda env remove -n ldm  # Use this if version 1 was previously installed
git clone https://github.com/Stability-AI/stablediffusion
cd stablediffusion
conda env create -f environment.yaml
conda activate ldm

Hugging Face offers a new weights ckpt file.

The Out of memory error prevented me from running this version on my 8 GB GPU. Version 2.1 fails on CPUs with the slow conv2d cpu not implemented for Half error (according to this GitHub issue, the CPU support for this algorithm and data type will not be added). The model can be modified from half to full precision (float16 instead of float32), however it doesn't make sense since v1 runs up to 10 minutes on the CPU and v2.1 should be much slower. The online demo results are visible. The same hamster painting with a brush prompt yielded this result:

It looks different from v1, but it functions and has a higher resolution.

The superresolution.py script can run the 4x Stable Diffusion upscaler locally (the x4-upscaler-ema.ckpt weights file should be in the same folder):

python3 scripts/gradio/superresolution.py configs/stable-diffusion/x4-upscaling.yaml x4-upscaler-ema.ckpt

This code allows the web browser UI to select the image to upscale:

The copy-paste strategy may explain why the upscaler needs a text prompt (and the Hugging Face code snippet does not have any text input as well). I got a GPU out of memory error again, although CUDA can be disabled like v1. However, processing an image for more than two hours is unlikely:

Stable Diffusion Limitations

When we use the model, it's fun to see what it can and can't do. Generative models produce abstract visuals but not photorealistic ones. This fundamentally limits The generative neural network was trained on text and image pairs, but humans have a lot of background knowledge about the world. The neural network model knows nothing. If someone asks me to draw a Chinese text, I can draw something that looks like Chinese but is actually gibberish because I never learnt it. Generative AI does too! Humans can learn new languages, but the Stable Diffusion AI model includes only language and image decoder brain components. For instance, the Stable Diffusion model will pull NO WAR banner-bearers like this:

V1:

V2.1:

The shot shows text, although the model never learned to read or write. The model's string tokenizer automatically converts letters to lowercase before generating the image, so typing NO WAR banner or no war banner is the same.

I can also ask the model to draw a gorgeous woman:

V1:

V2.1:

The first image is gorgeous but physically incorrect. A second one is better, although it has an Uncanny valley feel. BTW, v2 has a lifehack to add a negative prompt and define what we don't want on the image. Readers might try adding horrible anatomy to the gorgeous woman request.

If we ask for a cartoon attractive woman, the results are nice, but accuracy doesn't matter:

V1:

V2.1:

Another example: I ordered a model to sketch a mouse, which looks beautiful but has too many legs, ears, and fingers:

V1:

V2.1: improved but not perfect.

V1 produces a fun cartoon flying mouse if I want something more abstract:

I tried multiple times with V2.1 but only received this:

The image is OK, but the first version is closer to the request.

Stable Diffusion struggles to draw letters, fingers, etc. However, abstract images yield interesting outcomes. A rural landscape with a modern metropolis in the background turned out well:

V1:

V2.1:

Generative models help make paintings too (at least, abstract ones). I searched Google Image Search for modern art painting to see works by real artists, and this was the first image:

I typed "abstract oil painting of people dancing" and got this:

V1:

V2.1:

It's a different style, but I don't think the AI-generated graphics are worse than the human-drawn ones.

The AI model cannot think like humans. It thinks nothing. A stable diffusion model is a billion-parameter matrix trained on millions of text-image pairs. I input "robot is creating a picture with a pen" to create an image for this post. Humans understand requests immediately. I tried Stable Diffusion multiple times and got this:

This great artwork has a pen, robot, and sketch, however it was not asked. Maybe it was because the tokenizer deleted is and a words from a statement, but I tried other requests such robot painting picture with pen without success. It's harder to prompt a model than a person.

I hope Stable Diffusion's general effects are evident. Despite its limitations, it can produce beautiful photographs in some settings. Readers who want to use Stable Diffusion results should be warned. Source code examination demonstrates that Stable Diffusion images feature a concealed watermark (text StableDiffusionV1 and SDV2) encoded using the invisible-watermark Python package. It's not a secret, because the official Stable Diffusion repository's test watermark.py file contains a decoding snippet. The put watermark line in the txt2img.py source code can be removed if desired. I didn't discover this watermark on photographs made by the online Hugging Face demo. Maybe I did something incorrectly (but maybe they are just not using the txt2img script on their backend at all).

Conclusion

The Stable Diffusion model was fascinating. As I mentioned before, trying something yourself is always better than taking someone else's word, so I encourage readers to do the same (including this article as well;).

Is Generative AI a game-changer? My humble experience tells me:

• I think that place has a lot of potential. For designers and artists, generative AI can be a truly useful and innovative tool. Unfortunately, it can also pose a threat to some of them since if users can enter a text field to obtain a picture or a website logo in a matter of clicks, why would they pay more to a different party? Is it possible right now? unquestionably not yet. Images still have a very poor quality and are erroneous in minute details. And after viewing the image of the stunning woman above, models and fashion photographers may also unwind because it is highly unlikely that AI will replace them in the upcoming years.

• Today, generative AI is still in its infancy. Even 768x768 images are considered to be of a high resolution when using neural networks, which are computationally highly expensive. There isn't an AI model that can generate high-resolution photographs natively without upscaling or other methods, at least not as of the time this article was written, but it will happen eventually.

• It is still a challenge to accurately represent knowledge in neural networks (information like how many legs a cat has or the year Napoleon was born). Consequently, AI models struggle to create photorealistic photos, at least where little details are important (on the other side, when I searched Google for modern art paintings, the results are often even worse;).

• When compared to the carefully chosen images from official web pages or YouTube reviews, the average output quality of a Stable Diffusion generation process is actually less attractive because to its high degree of randomness. When using the same technique on their own, consumers will theoretically only view those images as 1% of the results.

Anyway, it's exciting to witness this area's advancement, especially because the project is open source. Google's Imagen and DALL-E 2 can also produce remarkable findings. It will be interesting to see how they progress.

I'm gadget-obsessed. So I shared my top 6 desk gadgets.

These gadgets improve my workflow and are handy for working from home.

Without further ado...

Computer light bar Xiaomi Mi

I've previously recommended the Xiaomi Mi Light Bar, and I still do. It's stylish and convenient.

The Mi bar is a monitor-mounted desk lamp. The lamp's hue and brightness can be changed with a stylish wireless remote.

Changeable hue and brightness make it ideal for late-night work.

Desk Mat 2.

I wasn't planning to include a desk surface in this article, but I find it improves computer use.

The mouse feels smoother and is a better palm rest than wood or glass.

I'm currently using the overkill Razer Goliathus Extended Chroma RGB Gaming Surface, but I like RGB.

Using a desk surface or mat makes computer use more comfortable, and it's not expensive.

Third, the Logitech MX Master 3 Mouse

The Logitech MX Master 3 or any from the MX Master series is my favorite mouse.

The side scroll wheel on these mice is a feature I've never seen on another mouse.

Side scroll wheels are great for spreadsheets and video editing. It would be hard for me to switch from my Logitech MX Master 3 to another mouse. Only gaming is off-limits.

Google Nest 4.

Without a smart assistant, my desk is useless. I'm currently using the second-generation Google Nest Hub, but I've also used the Amazon Echo Dot, Echo Spot, and Apple HomePod Mini.

As a Pixel 6 Pro user, the Nest Hub works best with my phone.

My Nest Hub plays news, music, and calendar events. It also lets me control lights and switches with my smartphone. It plays YouTube videos.

Google Pixel Stand, No. 5

A wireless charger on my desk is convenient for charging my phone and other devices while I work. My desk has two wireless chargers. I have a Satechi aluminum fast charger and a second-generation Google Pixel Stand.

If I need to charge my phone and earbuds simultaneously, I use two wireless chargers. Satechi chargers are well-made and fast. Micro-USB is my only complaint.

The Pixel Stand converts compatible devices into a smart display for adjusting charging speeds and controlling other smart devices. My Pixel 6 Pro charges quickly. Here's my video review.

6. Anker Power Bank

Anker's 65W charger is my final recommendation. This online find was a must-have. This can charge my laptop and several non-wireless devices, perfect for any techie!

The charger has two USB-A ports and two USB-C ports, one with 45W and the other with 20W, so it can charge my iPad Pro and Pixel 6 Pro simultaneously.

Summary

These are some of my favorite office gadgets. My kit page has an updated list.

Links to the products mentioned in this article are in the appropriate sections. These are affiliate links.

You're up! Share the one desk gadget you can't live without and why.

My kids redesigned the internet because it lacks inventiveness.

Internet today is bland. Everything is generic: fonts, layouts, pages, and visual language. Microtypography is messy.

Web design today seems dictated by technical and ideological constraints rather than creativity and ideas. Text and graphics are in containers on every page. All design is assumed.

Ironically, web technologies can design a lot. We can execute most designs. We make shocking, evocative websites. Experimental typography, generating graphics, and interactive experiences are possible.

Even designer websites use containers in containers. Dribbble and Behance, the two most popular creative websites, are boring. Lead image.

How did this happen?

Several reasons. WordPress and other blogging platforms use templates. These frameworks build web pages by combining graphics, headlines, body content, and videos. Not designs, templates. These rules combine related data types. These platforms don't let users customize pages beyond the template. You filled the template.

Templates are content-neutral. Thus, the issue.

Form should reflect and shape content, which is a design principle. Separating them produces content containers. Templates have no design value.

One of the fundamental principles of design is a deep and meaningful connection between form and content.

Web design lacks imagination for many reasons. Most are pragmatic and economic. Page design takes time. Large websites lack the resources to create a page from scratch due to the speed of internet news and the frequency of new items. HTML, JavaScript, and CSS continue to challenge web designers. Web design can't match desktop publishing's straightforward operations.

Designers may also be lazy. Mobile-first, generic, framework-driven development tends to ignore web page visual and contextual integrity.

How can we overcome this? How might expressive and avant-garde websites look today?

Rediscovering the past helps design the future.

'90s-era web design

At the University of the Arts Bremen's research and development group, I created my first website 23 years ago. Web design was trendy. Young web. Pages inspired me.

We struggled with HTML in the mid-1990s. Arial, Times, and Verdana were the only web-safe fonts. Anything exciting required table layouts, monospaced fonts, or GIFs. HTML was originally content-driven, thus we had to work against it to create a page.

Experimental typography was booming. Designers challenged the established quo from Jan Tschichold's Die Neue Typographie in the twenties to April Greiman's computer-driven layouts in the eighties. By the mid-1990s, an uncommon confluence of technological and cultural breakthroughs enabled radical graphic design. Irma Boom, David Carson, Paula Scher, Neville Brody, and others showed it.

Early web pages were dull compared to graphic design's aesthetic explosion. The Web Design Museum shows this.

Nobody knew how to conduct browser-based graphic design. Web page design was undefined. No standards. No CMS (nearly), CSS, JS, video, animation.

Now is as good a time as any to challenge the internet’s visual conformity.

In 2018, everything is browser-based. Massive layouts to micro-typography, animation, and video. How do we use these great possibilities? Containerized containers. JavaScript-contaminated mobile-first pages. Visually uniform templates. Web design 23 years later would disappoint my younger self.

Our imagination, not technology, restricts web design. We're too conformist to aesthetics, economics, and expectations.

Crisis generates opportunity. Challenge online visual conformity now. I'm too old and bourgeois to develop a radical, experimental, and cutting-edge website. I can ask my students.

I taught web design at the Potsdam Interface Design Programme in 2017. Each team has to redesign a website. Create expressive, inventive visual experiences on the browser. Create with contemporary web technologies. Avoid usability, readability, and flexibility concerns. Act. Ignore Erwartungskonformität.

The class outcome pleased me. This overview page shows all results. Four diverse projects address the challenge.

1. ZKM by Frederic Haase and Jonas Köpfer

Frederic and Jonas began their experiments on the ZKM website. The ZKM is Germany's leading media art exhibition location, but its website remains conventional. It's useful but not avant-garde like the shows' art.

Frederic and Jonas designed the ZKM site's concept, aesthetic language, and technical configuration to reflect the museum's progressive approach. A generative design engine generates new layouts for each page load.

ZKM redesign.

2. Streem by Daria Thies, Bela Kurek, and Lucas Vogel

Street art magazine Streem. It promotes new artists and societal topics. Streem includes artwork, painting, photography, design, writing, and journalism. Daria, Bela, and Lucas used these influences to develop a conceptual metropolis. They designed four neighborhoods to reflect magazine sections for their prototype. For a legible city, they use powerful illustrative styles and spatial typography.

Streem makeover.

3. Medium by Amelie Kirchmeyer and Fabian Schultz

Amelie and Fabian structured. Instead of developing a form for a tale, they dissolved a web page into semantic, syntactical, and statistical aspects. HTML's flexibility was their goal. They broke Medium posts into experimental typographic space.

Medium revamp.

4. Hacker News by Fabian Dinklage and Florian Zia

Florian and Fabian made Hacker News interactive. The social networking site aggregates computer science and IT news. Its voting and debate features are extensive despite its simple style. Fabian and Florian transformed the structure into a typographic timeline and network area. News and comments sequence and connect the visuals. To read Hacker News, they connected their design to the API. Hacker News makeover.

Communication is not legibility, said Carson. Apply this to web design today. Modern websites must be legible, usable, responsive, and accessible. They shouldn't limit its visual palette. Visual and human-centered design are not stereotypes.

I want radical, generative, evocative, insightful, adequate, content-specific, and intelligent site design. I want to rediscover web design experimentation. More surprises please. I hope the web will appear different in 23 years.

Update: this essay has sparked a lively discussion! I wrote a brief response to the debate's most common points: Creativity vs. Usability